1 00:00:00,000 --> 00:00:00,380 2 00:00:00,380 --> 00:00:01,040 GREG MITTLEIDER: Good evening, everybody. 3 00:00:01,040 --> 00:00:03,164 My name is Greg Mittleider, and on behalf of cs 50, 4 00:00:03,164 --> 00:00:07,700 I want to thank you once again for joining us for a TED Talk in our series 5 00:00:07,700 --> 00:00:09,920 this fall semester. 6 00:00:09,920 --> 00:00:13,590 Tonight we have the pleasure of having our friends from Uber with us. 7 00:00:13,590 --> 00:00:16,280 They're going to talk about the obstacles they 8 00:00:16,280 --> 00:00:19,730 have to overcome in what seems to be a very simple task by us, 9 00:00:19,730 --> 00:00:23,290 the user, which is how the car gets to you whenever 10 00:00:23,290 --> 00:00:25,430 you choose that you need a ride. 11 00:00:25,430 --> 00:00:28,314 Tonight we have the pleasure of having Matthew, Yuki, and Alli. 12 00:00:28,314 --> 00:00:30,230 They're going to be speaking with you tonight. 13 00:00:30,230 --> 00:00:32,890 So without further ado, Uber. 14 00:00:32,890 --> 00:00:35,210 YUKI: Thank you so much. 15 00:00:35,210 --> 00:00:36,560 Cool. 16 00:00:36,560 --> 00:00:41,767 so today, Matt and I are PMs at Uber, product managers at Uber. 17 00:00:41,767 --> 00:00:43,850 And we're here to talk about how we build products 18 00:00:43,850 --> 00:00:47,060 at Uber, how we bridge the physical and digital worlds. 19 00:00:47,060 --> 00:00:49,880 20 00:00:49,880 --> 00:00:53,305 At the expense of making us seem really old, 21 00:00:53,305 --> 00:00:56,390 we wanted to talk a little bit about what 22 00:00:56,390 --> 00:01:00,150 Harvard was like before the Uber days. 23 00:01:00,150 --> 00:01:04,760 Matt was concentrating in English and busy doing pre-med 24 00:01:04,760 --> 00:01:08,570 in the beautiful house with Elliot. 25 00:01:08,570 --> 00:01:12,660 And I was concentrating in computer science and TF and cs50 26 00:01:12,660 --> 00:01:15,809 in superior house that is Mather. 27 00:01:15,809 --> 00:01:17,100 MATTHEW: You can do the slides. 28 00:01:17,100 --> 00:01:20,030 YUKI: Yeah, I did the slides. 29 00:01:20,030 --> 00:01:22,310 And in terms of how we got around, of course, 30 00:01:22,310 --> 00:01:23,800 we used the T to get into the city. 31 00:01:23,800 --> 00:01:25,910 I think all of you still do that. 32 00:01:25,910 --> 00:01:32,450 But there's always the inconvenience of needing to come back before midnight. 33 00:01:32,450 --> 00:01:34,850 We took cabs to get to the airport. 34 00:01:34,850 --> 00:01:38,000 And of course, we took the Harvard shuttles to get around campus, 35 00:01:38,000 --> 00:01:40,270 the Matter express being my favorite. 36 00:01:40,270 --> 00:01:43,400 And Matt and I liked using this to visit our sad friends 37 00:01:43,400 --> 00:01:48,790 in the Quad who didn't quite make the cut for our blocking groups. 38 00:01:48,790 --> 00:01:51,890 But CS50, in terms of the innovation that 39 00:01:51,890 --> 00:01:54,020 was happening in this transportation space, 40 00:01:54,020 --> 00:01:58,760 CS50 was pretty eager to make the shuttle experience better. 41 00:01:58,760 --> 00:02:03,270 And David built Shuttle Boy back in 1999. 42 00:02:03,270 --> 00:02:06,710 And what we did was also take this massive PDF 43 00:02:06,710 --> 00:02:10,880 that people had to parse to understand settle schedules, 44 00:02:10,880 --> 00:02:13,027 and we have made these CS 50 shuttle cards, 45 00:02:13,027 --> 00:02:15,860 which are these tiny little things that you could put in the wallet. 46 00:02:15,860 --> 00:02:18,900 We printed thousands of them and distributed it across campus. 47 00:02:18,900 --> 00:02:21,702 So that was the kind of innovation that was happening at the time. 48 00:02:21,702 --> 00:02:23,030 MATTHEW: They were viral. 49 00:02:23,030 --> 00:02:27,170 YUKI: Yeah, they were kind of viral, but we clearly weren't thinking big enough. 50 00:02:27,170 --> 00:02:28,190 But it was really funny. 51 00:02:28,190 --> 00:02:30,660 I dug up some e-mails with David back in the day. 52 00:02:30,660 --> 00:02:32,905 I was really excited about these things. 53 00:02:32,905 --> 00:02:35,530 And so I would ping him late at night about the Matter schedule 54 00:02:35,530 --> 00:02:36,250 or something like that. 55 00:02:36,250 --> 00:02:38,240 So something I was really passionate about, 56 00:02:38,240 --> 00:02:43,764 but that was the kind of innovation that was happening on Harvard. 57 00:02:43,764 --> 00:02:46,280 MATTHEW: So that was kind of what working at Harvard 58 00:02:46,280 --> 00:02:47,830 was like back when we were in school. 59 00:02:47,830 --> 00:02:49,810 But obviously, things have changed quite a bit. 60 00:02:49,810 --> 00:02:51,160 We were in school six years ago. 61 00:02:51,160 --> 00:02:52,330 There was no Uber. 62 00:02:52,330 --> 00:02:54,080 Today that's not quite the case. 63 00:02:54,080 --> 00:02:57,790 So if you look now at a map of Harvard or Cambridge, 64 00:02:57,790 --> 00:03:00,270 Harvard Square, you can see each of these yellow dots 65 00:03:00,270 --> 00:03:02,380 represents an Uber trip that's begun. 66 00:03:02,380 --> 00:03:04,700 And this is just data from the last week. 67 00:03:04,700 --> 00:03:07,500 So we're moving quite a few people around the city from point A 68 00:03:07,500 --> 00:03:09,730 to point B, which is pretty cool. 69 00:03:09,730 --> 00:03:13,840 And if you zoom in a bit, and you just look at trips that begin or end at some 70 00:03:13,840 --> 00:03:16,810 of the Harvard houses-- this is also from like a weekend-- 71 00:03:16,810 --> 00:03:20,944 you see that we're doing some good work bridging the Quad, River House divide, 72 00:03:20,944 --> 00:03:23,860 which again, as someone who lived in Elliott, I think is really sweet. 73 00:03:23,860 --> 00:03:27,109 Because I honestly couldn't tell you the difference between Currier and Cabot. 74 00:03:27,109 --> 00:03:30,930 But apparently, we're making that world smaller, which is really wonderful. 75 00:03:30,930 --> 00:03:33,670 And if you just take a look at the count of trips by house, 76 00:03:33,670 --> 00:03:36,170 it's also really interesting to see some of the trends here. 77 00:03:36,170 --> 00:03:39,942 So there are a certain set of houses that people really like to visit, 78 00:03:39,942 --> 00:03:41,150 they're being dropped off at. 79 00:03:41,150 --> 00:03:43,640 And there are other houses that people seem to want to leave, 80 00:03:43,640 --> 00:03:44,470 and they're being picked up at. 81 00:03:44,470 --> 00:03:46,070 So this is kind of the relative ranking in terms 82 00:03:46,070 --> 00:03:47,611 of where people are going and coming. 83 00:03:47,611 --> 00:03:51,300 YUKI: We're a little bit unsure what's going on in Winthrop. 84 00:03:51,300 --> 00:03:55,340 MATTHEW: Yeah, and not [? happening. ?] And then Uber 85 00:03:55,340 --> 00:03:58,106 has another business called UberEATS, which is food deliveries. 86 00:03:58,106 --> 00:04:00,230 You can get food delivered from a local restaurant. 87 00:04:00,230 --> 00:04:05,150 If you look at the number of deliveries made per house, 88 00:04:05,150 --> 00:04:07,937 it's also kind of interesting to see which dining halls may not 89 00:04:07,937 --> 00:04:10,770 be serving the needs of their students quite as well at other homes. 90 00:04:10,770 --> 00:04:13,630 So [INAUDIBLE] what's happening at Mather, but maybe Yuki can-- 91 00:04:13,630 --> 00:04:17,300 92 00:04:17,300 --> 00:04:21,019 So Uber isn't popular just at Harvard, and hasn't changed transportation just 93 00:04:21,019 --> 00:04:22,190 here at Harvard. 94 00:04:22,190 --> 00:04:25,280 We are in tons of cities all over the world. 95 00:04:25,280 --> 00:04:29,630 And if you just look at in the US, these are our five biggest cities 96 00:04:29,630 --> 00:04:32,380 as we rolled out five years ago. 97 00:04:32,380 --> 00:04:34,710 And you can see there's lots of interesting data here. 98 00:04:34,710 --> 00:04:37,880 One is that our growth has been almost exponential, which is pretty cool. 99 00:04:37,880 --> 00:04:40,921 And if you look-- this is in the order of which we launched our bay city, 100 00:04:40,921 --> 00:04:44,516 so SF first, then New York, then Seattle, then Chicago, then DC. 101 00:04:44,516 --> 00:04:46,890 That growth rate is increasing as we roll out a new city. 102 00:04:46,890 --> 00:04:50,690 So as we launch in a new city, the adoption 103 00:04:50,690 --> 00:04:53,600 is even faster than it was in previous cities, which is pretty cool. 104 00:04:53,600 --> 00:04:55,433 And if you look at our trend globally, we're 105 00:04:55,433 --> 00:04:58,970 in tons of countries, which we'll get to in a minute. 106 00:04:58,970 --> 00:05:02,090 This growth is also really rapid. 107 00:05:02,090 --> 00:05:04,640 And we're now basically everywhere in the world you go, 108 00:05:04,640 --> 00:05:06,230 you can pretty much take an Uber. 109 00:05:06,230 --> 00:05:08,120 Latin America being our faster growing market right now. 110 00:05:08,120 --> 00:05:09,869 And I'm actually going to Colombia tonight 111 00:05:09,869 --> 00:05:12,140 to go see what's happening with our cities there, 112 00:05:12,140 --> 00:05:14,570 and see how we can make our product experience better for our users. 113 00:05:14,570 --> 00:05:16,444 YUKI: And for those of you who are observant, 114 00:05:16,444 --> 00:05:18,820 this is in a logarithmic scale, and so we 115 00:05:18,820 --> 00:05:22,980 had to adjust this chart to make it fit. 116 00:05:22,980 --> 00:05:26,910 MATTHEW: All right. so Uber today, 2017, we're at 600 plus cities. 117 00:05:26,910 --> 00:05:28,230 And we're like 660 right now. 118 00:05:28,230 --> 00:05:29,370 We're in 77 countries. 119 00:05:29,370 --> 00:05:32,760 And as of June 29, we hit our five billionth trip, 120 00:05:32,760 --> 00:05:33,914 which is pretty amazing. 121 00:05:33,914 --> 00:05:36,330 But the cool thing is these numbers are changing every day 122 00:05:36,330 --> 00:05:38,788 and growing everyday, or expanding to more and more cities. 123 00:05:38,788 --> 00:05:42,810 Like just the other week, we launched a new city in Croatia 124 00:05:42,810 --> 00:05:44,040 on some small islands. 125 00:05:44,040 --> 00:05:45,405 And these Uber's were all boats. 126 00:05:45,405 --> 00:05:47,282 So they were contacting us, and being like, 127 00:05:47,282 --> 00:05:48,990 can you change the app to work for boats? 128 00:05:48,990 --> 00:05:50,930 And I was like, OK, we can do that, I guess. 129 00:05:50,930 --> 00:05:55,110 So there's a lot of land still to be grabbed. 130 00:05:55,110 --> 00:05:57,771 And it's really exciting times for us. 131 00:05:57,771 --> 00:06:00,210 YUKI: Cool, so this all sounds really simple. 132 00:06:00,210 --> 00:06:02,430 And we're doing billions of rides. 133 00:06:02,430 --> 00:06:06,420 And it seems really simple to get someone from point A to point B. 134 00:06:06,420 --> 00:06:09,600 But it turns out that a lot can go wrong. 135 00:06:09,600 --> 00:06:11,790 And specifically, the thing that can go wrong 136 00:06:11,790 --> 00:06:15,660 is [? always ?] alluded to earlier, which is actually 137 00:06:15,660 --> 00:06:17,380 getting picked up by your driver. 138 00:06:17,380 --> 00:06:22,020 And so if you think about the journey for a rider, you open the app, 139 00:06:22,020 --> 00:06:24,390 you request a car, and you get into the car. 140 00:06:24,390 --> 00:06:27,750 But actually, this particular part is the most difficult part. 141 00:06:27,750 --> 00:06:30,810 And this is where we need to bridge the physical and digital worlds. 142 00:06:30,810 --> 00:06:33,840 So I'm going to first talk a little bit about this period 143 00:06:33,840 --> 00:06:37,290 right here, which is just the period between the opening up of the app 144 00:06:37,290 --> 00:06:45,220 and requesting a car and all the tech that goes into making this happen. 145 00:06:45,220 --> 00:06:49,357 So the first question we ask when you open up the app is, where is the rider? 146 00:06:49,357 --> 00:06:52,440 And this sounds like a really, really simple question, but figuring it out 147 00:06:52,440 --> 00:06:54,070 isn't always easy. 148 00:06:54,070 --> 00:06:58,660 And one of the reasons for this is because GPS is often inaccurate. 149 00:06:58,660 --> 00:07:01,245 And so there are many reasons that can cause this. 150 00:07:01,245 --> 00:07:04,650 This can be because there are tall buildings surrounding you, 151 00:07:04,650 --> 00:07:07,680 and they block the line of sight with the GPS satellites. 152 00:07:07,680 --> 00:07:11,160 There can also be buildings that cause reflections that misguide us. 153 00:07:11,160 --> 00:07:14,990 And therefore, I think all of us have had this experience where we open up 154 00:07:14,990 --> 00:07:17,196 our app-- whether it's the Uber app, Google Maps-- 155 00:07:17,196 --> 00:07:19,070 and the blue dot isn't really where we're at. 156 00:07:19,070 --> 00:07:22,880 And these are the kinds of problems that can produce this. 157 00:07:22,880 --> 00:07:25,970 We have another kind of problem, which is that reverse geocoding can 158 00:07:25,970 --> 00:07:26,720 be inaccurate. 159 00:07:26,720 --> 00:07:30,049 Reverse geocoding is the act of taking a geocode, which is the lat, lng, 160 00:07:30,049 --> 00:07:32,840 these coordinates that represent where you are, and then turning it 161 00:07:32,840 --> 00:07:33,890 into an address. 162 00:07:33,890 --> 00:07:36,920 So these are a lat lngs for this space. 163 00:07:36,920 --> 00:07:41,120 And when you reverse geocode it, hopefully what it produces is 67 Mt. 164 00:07:41,120 --> 00:07:42,230 Auburn street. 165 00:07:42,230 --> 00:07:45,920 But actually, if you look at some of the examples I'm going to show you, 166 00:07:45,920 --> 00:07:48,010 this can also be an inaccurate process. 167 00:07:48,010 --> 00:07:51,495 So as an example, I'm not sure if any of you have been to Tea Luxe 168 00:07:51,495 --> 00:07:52,120 in this square. 169 00:07:52,120 --> 00:07:55,760 If you have, if you happen to put your pin right in the middle of it, 170 00:07:55,760 --> 00:07:59,170 then it produces the correct address, which is 0 Brattle street. 171 00:07:59,170 --> 00:08:03,200 It turns out Brattle street is zero index, which is pretty cool. 172 00:08:03,200 --> 00:08:04,400 And so that's correct. 173 00:08:04,400 --> 00:08:09,080 But if you move your pin slightly away from it, then all of a sudden, 174 00:08:09,080 --> 00:08:10,850 we think you're in 1 JFK street. 175 00:08:10,850 --> 00:08:13,100 And for whatever reason, it may be because the outline 176 00:08:13,100 --> 00:08:14,582 of the POI, Tea Luxe-- 177 00:08:14,582 --> 00:08:16,790 the point of interest-- might be slightly inaccurate. 178 00:08:16,790 --> 00:08:20,567 It might be because it's actually a little bit closer to JFK. 179 00:08:20,567 --> 00:08:23,150 And this slight difference produces a different kind of pickup 180 00:08:23,150 --> 00:08:25,700 because the driver is going to show up on JFK street, which 181 00:08:25,700 --> 00:08:27,866 is very different than showing up on Brattle street, 182 00:08:27,866 --> 00:08:30,560 especially if you consider Tea Luxe, which I think only 183 00:08:30,560 --> 00:08:33,461 has an exit on Brattle street. 184 00:08:33,461 --> 00:08:35,419 And there are many kinds of examples like this. 185 00:08:35,419 --> 00:08:39,890 So this is the Cambridge Side Galleria for those of you who've been. 186 00:08:39,890 --> 00:08:43,502 So if you put a pin on top of PF Chang's, which 187 00:08:43,502 --> 00:08:44,960 is a Dave and [INAUDIBLE] favorite. 188 00:08:44,960 --> 00:08:47,450 I don't know if it's changed. 189 00:08:47,450 --> 00:08:52,130 Then what happens is we have a selection of addresses to choose from. 190 00:08:52,130 --> 00:08:53,380 It could be PF Chang's. 191 00:08:53,380 --> 00:08:55,880 You're technically still inside the Cambridge Side Galleria, 192 00:08:55,880 --> 00:08:58,190 so we could choose to show that instead. 193 00:08:58,190 --> 00:09:00,980 We could also show the street address for Cambridge Side Galleria. 194 00:09:00,980 --> 00:09:03,170 We could show the street address for PF Chang's. 195 00:09:03,170 --> 00:09:05,930 And you can imagine if we pick any one of these, 196 00:09:05,930 --> 00:09:08,630 it might produce a different kind of pickup. 197 00:09:08,630 --> 00:09:10,730 And so this is our fundamental problem, which 198 00:09:10,730 --> 00:09:14,840 is that your input, which is the lt lng, could be inaccurate. 199 00:09:14,840 --> 00:09:17,630 And then the process of translating that into an address 200 00:09:17,630 --> 00:09:18,730 could also be inaccurate. 201 00:09:18,730 --> 00:09:20,660 So your output is also erroneous. 202 00:09:20,660 --> 00:09:22,640 So this is what we're dealing with, which 203 00:09:22,640 --> 00:09:26,750 is a difficult situation to be in. 204 00:09:26,750 --> 00:09:29,060 So what do we do? 205 00:09:29,060 --> 00:09:31,160 Some of the ways we try to solve this is actually 206 00:09:31,160 --> 00:09:35,069 trying to predict where you are based on the information that we have. 207 00:09:35,069 --> 00:09:36,860 So if you really think through the problem, 208 00:09:36,860 --> 00:09:38,550 we started with some hypotheses. 209 00:09:38,550 --> 00:09:41,750 So for example, let's say that you recently 210 00:09:41,750 --> 00:09:44,870 got dropped off at Currier House. 211 00:09:44,870 --> 00:09:49,790 And even if the GPS is telling us that you're in Pfoho, 212 00:09:49,790 --> 00:09:51,800 we might actually believe that you are more 213 00:09:51,800 --> 00:09:54,680 likely to be in Currier because you just got dropped off there. 214 00:09:54,680 --> 00:09:56,560 So that could be an interesting heuristic 215 00:09:56,560 --> 00:09:58,940 to start using in terms of round trips. 216 00:09:58,940 --> 00:10:02,430 Or another hypothesis, and this is a simpler example. 217 00:10:02,430 --> 00:10:05,870 If you've saved a location-- so our app has the feature 218 00:10:05,870 --> 00:10:08,180 of basically favoring a location. 219 00:10:08,180 --> 00:10:11,720 So Mather House is a saved location obviously for me, both in the app 220 00:10:11,720 --> 00:10:13,170 and in my heart. 221 00:10:13,170 --> 00:10:18,950 And even though the GPS might be telling you we're not there, 222 00:10:18,950 --> 00:10:22,340 it's very likely because you saved it that you're 223 00:10:22,340 --> 00:10:24,430 going to be in this location or you care about it. 224 00:10:24,430 --> 00:10:26,560 So that's a signal that we can use. 225 00:10:26,560 --> 00:10:29,220 And when you start thinking about these, these hypotheses 226 00:10:29,220 --> 00:10:31,500 turn into what we call features. 227 00:10:31,500 --> 00:10:34,047 Features that inform our pick up prediction models. 228 00:10:34,047 --> 00:10:35,880 So our pickup prediction model takes a bunch 229 00:10:35,880 --> 00:10:38,640 of these different kind of parameters, if you will, and basically 230 00:10:38,640 --> 00:10:40,830 starts to build models on top of it. 231 00:10:40,830 --> 00:10:43,130 So other features include trip counts. 232 00:10:43,130 --> 00:10:46,427 So, for example, I worked at the Harvard Crimson. 233 00:10:46,427 --> 00:10:47,760 And I spent a lot of time there. 234 00:10:47,760 --> 00:10:52,920 So this is the hypothetical me, because these trips I never took. 235 00:10:52,920 --> 00:10:54,969 And I spent very little time at Adams House 236 00:10:54,969 --> 00:10:56,510 because they didn't let me eat there. 237 00:10:56,510 --> 00:10:58,390 I don't know if that's still a thing. 238 00:10:58,390 --> 00:11:01,157 But in any case, if the GP tells us that we 239 00:11:01,157 --> 00:11:03,240 think you're somewhere in between these buildings, 240 00:11:03,240 --> 00:11:06,900 we might actually favor the Crimson because you take a lot of trips there 241 00:11:06,900 --> 00:11:10,640 or you take a lot of trips from there. 242 00:11:10,640 --> 00:11:12,390 Another feature might be destination. 243 00:11:12,390 --> 00:11:15,650 So this is a slightly contrived example, but you 244 00:11:15,650 --> 00:11:18,502 have this area right near Lamont where there's 245 00:11:18,502 --> 00:11:20,210 the Department of Comparative Literature, 246 00:11:20,210 --> 00:11:21,875 and then there's the Barker Center. 247 00:11:21,875 --> 00:11:24,500 And depending on where you're going, if you're going northbound 248 00:11:24,500 --> 00:11:27,650 because these are both one way streets, we might favor the Barker Center. 249 00:11:27,650 --> 00:11:31,700 If you're going southbound, we might favor the other place. 250 00:11:31,700 --> 00:11:34,700 And another feature that's really interesting is time of day and day 251 00:11:34,700 --> 00:11:35,480 of week. 252 00:11:35,480 --> 00:11:41,870 So in the weekdays, you might be going to CS50 lectures 253 00:11:41,870 --> 00:11:42,960 or whatever that may be. 254 00:11:42,960 --> 00:11:46,950 And so if your GPS tells us that you're in an MBRD, 255 00:11:46,950 --> 00:11:50,540 we might actually intuit that you're actually closer to Sanders. 256 00:11:50,540 --> 00:11:53,750 But maybe in the evenings, or on Saturdays or something like that, 257 00:11:53,750 --> 00:11:55,430 you're in Queen's Head. 258 00:11:55,430 --> 00:12:01,120 So all these contribute to this model of figuring out where exactly you are. 259 00:12:01,120 --> 00:12:02,870 And we're experimenting with a lot of them 260 00:12:02,870 --> 00:12:04,703 all the time to make it smarter and smarter. 261 00:12:04,703 --> 00:12:07,924 262 00:12:07,924 --> 00:12:09,840 MATTHEW: So we talked a lot about figuring out 263 00:12:09,840 --> 00:12:12,840 where the rider is when they open the app, which is pretty awesome. 264 00:12:12,840 --> 00:12:15,881 But in order to make a trip actually happen, we need to get a car to you. 265 00:12:15,881 --> 00:12:16,950 So I work on routing. 266 00:12:16,950 --> 00:12:19,590 And one of the big problems is that where humans are 267 00:12:19,590 --> 00:12:22,230 isn't necessarily a place where a car can actually reach you. 268 00:12:22,230 --> 00:12:26,834 So we have this problem whee even if we know where you are, 269 00:12:26,834 --> 00:12:29,500 that doesn't mean it's the best place to actually get picked up. 270 00:12:29,500 --> 00:12:31,450 So we need to translate the place where you 271 00:12:31,450 --> 00:12:34,450 are, which we've decided now-- where we've helped to intuit-- to a place 272 00:12:34,450 --> 00:12:35,616 where a car can actually go. 273 00:12:35,616 --> 00:12:39,020 274 00:12:39,020 --> 00:12:40,489 So let's take an example. 275 00:12:40,489 --> 00:12:41,280 This is an airport. 276 00:12:41,280 --> 00:12:42,410 This is actually the airport in Dallas. 277 00:12:42,410 --> 00:12:44,326 And this is an image taken from two years ago. 278 00:12:44,326 --> 00:12:47,380 What this is depicting is a bunch of trips where every trip is 279 00:12:47,380 --> 00:12:49,660 represented by two points and a line. 280 00:12:49,660 --> 00:12:54,307 One point is the location where the rider was requesting to get picked up. 281 00:12:54,307 --> 00:12:55,390 So they get off the plane. 282 00:12:55,390 --> 00:12:57,880 They're walking to the food court, and they're requesting to get picked up. 283 00:12:57,880 --> 00:12:58,360 I wand to get pick up at the airport. 284 00:12:58,360 --> 00:13:01,274 The second thought is where the trip actually began. 285 00:13:01,274 --> 00:13:03,190 So what you can see is that people were asking 286 00:13:03,190 --> 00:13:04,960 to get picked up all over the airport-- 287 00:13:04,960 --> 00:13:07,480 at the food court, where they're picking up their luggage. 288 00:13:07,480 --> 00:13:10,180 Some people were even requesting a car from in the plane 289 00:13:10,180 --> 00:13:11,740 on the tarmac, which is impossible. 290 00:13:11,740 --> 00:13:13,551 Cars can't go to any of those places. 291 00:13:13,551 --> 00:13:15,550 So there is this coordination that has to happen 292 00:13:15,550 --> 00:13:17,800 where the driver has to call you, and be like, where 293 00:13:17,800 --> 00:13:19,000 do you actually want to get picked up? 294 00:13:19,000 --> 00:13:19,990 I can't go to the food court. 295 00:13:19,990 --> 00:13:22,420 And the rider has to be like, oh, I guess somewhere near departures, 296 00:13:22,420 --> 00:13:24,628 or maybe I'm not from here, I don't know where to go. 297 00:13:24,628 --> 00:13:25,832 It's very messy, right? 298 00:13:25,832 --> 00:13:27,790 But you'll see that most of the actual pick ups 299 00:13:27,790 --> 00:13:32,800 will happen somewhere along this road, which is where arrivals are, 300 00:13:32,800 --> 00:13:35,660 and you can actually get a car into this space. 301 00:13:35,660 --> 00:13:38,170 So what we did is we said, OK, this is problematic, 302 00:13:38,170 --> 00:13:39,550 and there's a team in Dallas. 303 00:13:39,550 --> 00:13:42,820 Every Uber city has a operations team that works there and they're local. 304 00:13:42,820 --> 00:13:44,920 We go to them and we say, hey, look, you guys live in Dallas. 305 00:13:44,920 --> 00:13:45,640 You know your airport. 306 00:13:45,640 --> 00:13:46,480 We don't know you airport. 307 00:13:46,480 --> 00:13:49,720 Can you tell us where at the Dallas airport people should get picked up? 308 00:13:49,720 --> 00:13:50,740 And they're like, yes, of course we can. 309 00:13:50,740 --> 00:13:51,980 We fly here all the time. 310 00:13:51,980 --> 00:13:53,021 These are the five spots. 311 00:13:53,021 --> 00:13:56,170 And so what we do is we actually mapped out all the places at the airport 312 00:13:56,170 --> 00:13:57,550 that someone can get picked up. 313 00:13:57,550 --> 00:14:00,440 And actually we indexed it by terminal, which is pretty cool. 314 00:14:00,440 --> 00:14:01,810 And so we launched this. 315 00:14:01,810 --> 00:14:03,850 And what this did was it cleaned up the space 316 00:14:03,850 --> 00:14:04,960 where people were actually requesting. 317 00:14:04,960 --> 00:14:08,001 So now when you land at the airport, we figure out you are at the airport 318 00:14:08,001 --> 00:14:09,610 using what you just described. 319 00:14:09,610 --> 00:14:12,340 And then we say, OK, you can only get picked up and request 320 00:14:12,340 --> 00:14:13,990 a ride from one of the select places. 321 00:14:13,990 --> 00:14:15,700 And that simplifies the problem for the drivers, 322 00:14:15,700 --> 00:14:17,170 so they know exactly where to go, and also 323 00:14:17,170 --> 00:14:19,295 for riders, because they know exactly where they're 324 00:14:19,295 --> 00:14:21,620 going to end up meeting their driver. 325 00:14:21,620 --> 00:14:22,750 So this works for airports. 326 00:14:22,750 --> 00:14:23,410 That's great. 327 00:14:23,410 --> 00:14:26,414 But how can we make this model work outside of airports? 328 00:14:26,414 --> 00:14:29,080 There's lots of places that are also difficult to get picked up. 329 00:14:29,080 --> 00:14:30,940 We wanted to scale this globally. 330 00:14:30,940 --> 00:14:32,982 So one idea is we could just go to the city teams 331 00:14:32,982 --> 00:14:35,481 and be like, hey, can you go to every building in your city, 332 00:14:35,481 --> 00:14:37,850 and just tell us exactly where you should get picked up. 333 00:14:37,850 --> 00:14:40,891 And the answer would probably be no, because that would be a ton of work. 334 00:14:40,891 --> 00:14:44,139 And that's super crazy, and no one human knows what every pickup 335 00:14:44,139 --> 00:14:45,430 location is for every building. 336 00:14:45,430 --> 00:14:48,010 YUKI: You could [INAUDIBLE] 337 00:14:48,010 --> 00:14:49,770 MATTHEW: We could do that actually. 338 00:14:49,770 --> 00:14:50,270 Good idea. 339 00:14:50,270 --> 00:14:52,880 Maybe we'll circle back afterwards. 340 00:14:52,880 --> 00:14:54,460 So that's probably not going to work. 341 00:14:54,460 --> 00:15:02,080 So what we also could do is we could just take the address of a given place, 342 00:15:02,080 --> 00:15:04,150 and move the pen to that road. 343 00:15:04,150 --> 00:15:06,870 So, for example, Leverett House is located on 28 DeWolfe street. 344 00:15:06,870 --> 00:15:08,703 You could say, OK, for Leverett House, let's 345 00:15:08,703 --> 00:15:10,900 just find the closest point on DeWolfe street 346 00:15:10,900 --> 00:15:12,610 and just have our pick ups happen there. 347 00:15:12,610 --> 00:15:15,401 And for a lot, this actually would be a pretty decent pick up spot. 348 00:15:15,401 --> 00:15:16,720 DeWolfe's a pretty chill space. 349 00:15:16,720 --> 00:15:17,511 There's a sidewalk. 350 00:15:17,511 --> 00:15:18,660 It's pretty good. 351 00:15:18,660 --> 00:15:19,880 But this sometimes breaks. 352 00:15:19,880 --> 00:15:20,930 So you look at Grendel's. 353 00:15:20,930 --> 00:15:22,010 Grendel's is a bar on the square. 354 00:15:22,010 --> 00:15:23,110 It's just down the street. 355 00:15:23,110 --> 00:15:25,150 May have been there if you're over 21. 356 00:15:25,150 --> 00:15:26,890 And one of the problems with Grendel's is 357 00:15:26,890 --> 00:15:30,760 that their address is on 89 Winthrop street, which is this road right here. 358 00:15:30,760 --> 00:15:33,462 But Winthrop street at this point is not car friendly. 359 00:15:33,462 --> 00:15:34,670 It's actually a walking path. 360 00:15:34,670 --> 00:15:37,412 So if we start sending cars to 89 Winthrop street, 361 00:15:37,412 --> 00:15:39,370 we're going to start running over people, which 362 00:15:39,370 --> 00:15:41,390 is really bad for people and for Uber. 363 00:15:41,390 --> 00:15:43,790 And I'd probably get fired. 364 00:15:43,790 --> 00:15:45,907 So we needed something else that works. 365 00:15:45,907 --> 00:15:47,740 Well, the thing that we have at our disposal 366 00:15:47,740 --> 00:15:50,390 is we have all this GPS data from all these trips that are happening. 367 00:15:50,390 --> 00:15:52,280 So we can look at all of our trips and say, 368 00:15:52,280 --> 00:15:54,970 let's intuit based on where people tend to get picked up, 369 00:15:54,970 --> 00:15:56,110 what's a good location? 370 00:15:56,110 --> 00:15:58,960 And if you do live at Leverett-- this is Leverett-- 371 00:15:58,960 --> 00:16:00,130 it actually looks OK. 372 00:16:00,130 --> 00:16:02,797 You can see there's some clusters here around the side entrance. 373 00:16:02,797 --> 00:16:04,630 There's a cluster here by the main entrance. 374 00:16:04,630 --> 00:16:05,500 It's decent. 375 00:16:05,500 --> 00:16:08,690 You might be able to intuit something from this. 376 00:16:08,690 --> 00:16:10,880 But you go back to Grendel's, our problem case, 377 00:16:10,880 --> 00:16:13,419 and if you plot the same data-- this is all from week-- 378 00:16:13,419 --> 00:16:15,710 there's just so many trips happening here on Grendel's. 379 00:16:15,710 --> 00:16:16,751 It's such a popular area. 380 00:16:16,751 --> 00:16:19,380 There's a ton of popular bars and restaurants there. 381 00:16:19,380 --> 00:16:22,680 It's kind of hard to gather signal from all of this. 382 00:16:22,680 --> 00:16:24,860 So we tried another approach. 383 00:16:24,860 --> 00:16:28,140 We look and we filter points only by trips that we're requested from 384 00:16:28,140 --> 00:16:28,640 Grendel's. 385 00:16:28,640 --> 00:16:31,715 So as before, we were looking at all the trips that began around Grendel's, we 386 00:16:31,715 --> 00:16:35,030 were only looking at trips of the person was at Grendel's when they requested. 387 00:16:35,030 --> 00:16:36,390 When you do that, the data is much cleaner. 388 00:16:36,390 --> 00:16:39,500 There's a much clearer cluster just in front of the bar, which is probably 389 00:16:39,500 --> 00:16:41,040 where you want to get picked up. 390 00:16:41,040 --> 00:16:44,005 And we can actually generate a point for the pick up right here. 391 00:16:44,005 --> 00:16:46,880 And so if you use the app today, if you got out to Grendel's tonight, 392 00:16:46,880 --> 00:16:49,820 have a beer and request an Uber home, you will see these points. 393 00:16:49,820 --> 00:16:55,200 And they're available for all types of locations all over the world. 394 00:16:55,200 --> 00:16:57,770 We go back to Lev, run the same filter. 395 00:16:57,770 --> 00:16:59,150 It also simplifies the data. 396 00:16:59,150 --> 00:17:00,567 Generates a much cleaner cluster. 397 00:17:00,567 --> 00:17:03,650 And if you see you in [INAUDIBLE] today or in [INAUDIBLE] when you open up 398 00:17:03,650 --> 00:17:07,280 the app, you'll actually see we found spots that are 399 00:17:07,280 --> 00:17:08,869 in alignment with Leveretts's doors. 400 00:17:08,869 --> 00:17:09,800 So it's actually pretty cool. 401 00:17:09,800 --> 00:17:12,230 We went walking around the other day because we hadn't been back 402 00:17:12,230 --> 00:17:13,130 since this was launched. 403 00:17:13,130 --> 00:17:16,339 And we were like, oh, my gosh, our spots are at the door at Leverett, which 404 00:17:16,339 --> 00:17:17,000 is so cool. 405 00:17:17,000 --> 00:17:18,680 We didn't even tell the system. 406 00:17:18,680 --> 00:17:21,138 It just learned that based on the trips that were naturally 407 00:17:21,138 --> 00:17:23,359 occurring there, which is pretty cool. 408 00:17:23,359 --> 00:17:26,561 Now this approach is actually really important for what we call nested POIs. 409 00:17:26,561 --> 00:17:28,810 And these are really common, especially in big cities. 410 00:17:28,810 --> 00:17:31,970 What this is, essentially it's a building 411 00:17:31,970 --> 00:17:35,580 that has multiple places within that building. 412 00:17:35,580 --> 00:17:38,637 So for example, the building that we're in has multiple offices. 413 00:17:38,637 --> 00:17:40,220 The example here is the Charles Hotel. 414 00:17:40,220 --> 00:17:42,200 The Charles Hotel is a very fancy hotel in Cambridge. 415 00:17:42,200 --> 00:17:43,616 Maybe some of you have been there. 416 00:17:43,616 --> 00:17:44,990 Your parents have stayed there. 417 00:17:44,990 --> 00:17:46,000 It's a big complex. 418 00:17:46,000 --> 00:17:47,750 And within that complex is the main hotel. 419 00:17:47,750 --> 00:17:49,958 And these are all the requests points for that hotel. 420 00:17:49,958 --> 00:17:53,429 And you can see that most people were requesting at the front of the hotel. 421 00:17:53,429 --> 00:17:55,220 And most the trips, which are in green, are 422 00:17:55,220 --> 00:17:56,290 beginning at the front of the hotel. 423 00:17:56,290 --> 00:17:59,331 There's a massive cluster here that we're showing you, which makes sense. 424 00:17:59,331 --> 00:18:02,570 Most trips at the hotel will happen at the front door of the hotel. 425 00:18:02,570 --> 00:18:06,200 But also in this complex is Legal Sea Foods, 426 00:18:06,200 --> 00:18:09,260 which is a little seafood restaurant, pretty fancy. 427 00:18:09,260 --> 00:18:10,427 YUKI: Also a David favorite. 428 00:18:10,427 --> 00:18:12,218 MATTHEW: Also a David [INAUDIBLE] favorite. 429 00:18:12,218 --> 00:18:14,180 Your professor's might take you there sometime. 430 00:18:14,180 --> 00:18:16,730 And it's kind of nestled in the back of this complex. 431 00:18:16,730 --> 00:18:21,070 Well, if you look at the trips that start from this location 432 00:18:21,070 --> 00:18:24,809 where the requests happen, they mostly start over in this section. 433 00:18:24,809 --> 00:18:26,600 So you probably want to have a pick up spot 434 00:18:26,600 --> 00:18:28,599 over on university road, which is actually where 435 00:18:28,599 --> 00:18:30,260 the front door of Legal Sea Foods is. 436 00:18:30,260 --> 00:18:34,310 But if you use both sets of data, the number of trips at the Charles 437 00:18:34,310 --> 00:18:36,845 is a much bigger location, much more popular spot. 438 00:18:36,845 --> 00:18:39,470 It's going to totally outweigh what happens at Legal Sea Foods. 439 00:18:39,470 --> 00:18:41,010 So you're probably only going to generate one spot, 440 00:18:41,010 --> 00:18:42,820 and it's going to get the Charles, which is bad 441 00:18:42,820 --> 00:18:44,778 if you're eating at Legal Sea Foods, and you're 442 00:18:44,778 --> 00:18:47,530 old and can't walk through the whole hotel to get there. 443 00:18:47,530 --> 00:18:49,390 It's just super unideal. 444 00:18:49,390 --> 00:18:51,140 So by filtering by request location, we're 445 00:18:51,140 --> 00:18:55,372 actually able to generate different points for each spot. 446 00:18:55,372 --> 00:18:57,330 So this is actually a sample from the live app. 447 00:18:57,330 --> 00:18:58,560 If you're at Legal Sea Foods, we're going 448 00:18:58,560 --> 00:18:59,690 to send you to University, which is great 449 00:18:59,690 --> 00:19:01,010 because that's where the front door is. 450 00:19:01,010 --> 00:19:02,150 And if you're at the Charles Hotel, we're 451 00:19:02,150 --> 00:19:04,010 going to send you to Bennett street because that's 452 00:19:04,010 --> 00:19:05,093 where their front door is. 453 00:19:05,093 --> 00:19:07,300 So essentially we're able to crowdsource and learn 454 00:19:07,300 --> 00:19:10,695 what are the entrances and exits are based on user behavior. 455 00:19:10,695 --> 00:19:13,820 So that being said, there are so many ways we can make these pick up points 456 00:19:13,820 --> 00:19:15,290 better and we're working on. 457 00:19:15,290 --> 00:19:17,081 One of the things I think a lot about right 458 00:19:17,081 --> 00:19:19,890 now is how to refresh these and keep these new and up-to-date. 459 00:19:19,890 --> 00:19:24,310 Because as the real world changes, we have to be dynamic and change with it. 460 00:19:24,310 --> 00:19:27,767 So as road construction happens or a building goes under remodeling, 461 00:19:27,767 --> 00:19:30,725 we need to be able to adjust quickly, so that our riders, too, can meet 462 00:19:30,725 --> 00:19:33,850 their drivers in a reasonable place. 463 00:19:33,850 --> 00:19:36,190 YUKI: So we just talked just about the space 464 00:19:36,190 --> 00:19:39,310 of that time between opening up, requesting a car, 465 00:19:39,310 --> 00:19:40,699 and all the things that happened. 466 00:19:40,699 --> 00:19:42,490 But as you can imagine beyond that, there's 467 00:19:42,490 --> 00:19:43,880 a lot more that needs to happen. 468 00:19:43,880 --> 00:19:46,930 So once you request a car, everything that happens in between that 469 00:19:46,930 --> 00:19:51,360 and you getting in a car, there's many, many more problems. 470 00:19:51,360 --> 00:19:53,950 So as an example just to go through some of them quickly, 471 00:19:53,950 --> 00:19:58,330 how do we compute accurate ETAs that are aware of real-time traffic? 472 00:19:58,330 --> 00:20:00,400 This is a thing that actually annoys riders 473 00:20:00,400 --> 00:20:05,067 the most when ETAs are either wrong or they fluctuate. 474 00:20:05,067 --> 00:20:08,150 We have another problem of making sure that the car positions are actually 475 00:20:08,150 --> 00:20:09,290 accurate and in real-time. 476 00:20:09,290 --> 00:20:10,998 And there's a lot of interesting problems 477 00:20:10,998 --> 00:20:14,090 around how to deal with stale data when we lose connectivity or the driver 478 00:20:14,090 --> 00:20:16,940 loses connectivity. 479 00:20:16,940 --> 00:20:20,920 There's another question of how we get riders to be at the curb 480 00:20:20,920 --> 00:20:23,050 exactly when the driver arrives. 481 00:20:23,050 --> 00:20:28,780 And so we have a class of some lazy riders who don't actually 482 00:20:28,780 --> 00:20:32,185 leave their dorms or leave their rooms until the car actually arrives. 483 00:20:32,185 --> 00:20:34,060 And that produces a different kind of problem 484 00:20:34,060 --> 00:20:37,590 because the driver has to pull over. 485 00:20:37,590 --> 00:20:40,180 And there's an entire side of the driver experience 486 00:20:40,180 --> 00:20:41,620 that we haven't even touched. 487 00:20:41,620 --> 00:20:43,930 And I now work on the driver experience. 488 00:20:43,930 --> 00:20:46,070 And this is definitely the most stressful thing. 489 00:20:46,070 --> 00:20:51,490 It's much more stressful for the driver to actually get a rider picked up. 490 00:20:51,490 --> 00:20:54,640 And then there's this last 50 meter problem 491 00:20:54,640 --> 00:20:56,950 where riders and drivers might not be able to find 492 00:20:56,950 --> 00:21:01,150 each other because it's a really dark area, or because it's really crowded. 493 00:21:01,150 --> 00:21:04,600 And so as an example, this is where we've developed technology 494 00:21:04,600 --> 00:21:07,949 like this where a rider can choose a color of their liking, 495 00:21:07,949 --> 00:21:10,240 and a light that's attached to the windshield, which we 496 00:21:10,240 --> 00:21:12,200 call a beacon, lights up in that way. 497 00:21:12,200 --> 00:21:17,770 So in a dark venue, or when it's really crowded, this might be helpful. 498 00:21:17,770 --> 00:21:20,637 But there are a lot of these different problems around the pick ups. 499 00:21:20,637 --> 00:21:22,720 And there's a lot more problems outside of it too. 500 00:21:22,720 --> 00:21:26,800 501 00:21:26,800 --> 00:21:32,111 We're going to stay and go through some Q&A at the end, 502 00:21:32,111 --> 00:21:34,360 but really encourage you to ask us a lot of questions, 503 00:21:34,360 --> 00:21:36,320 and come work with us on these problems. 504 00:21:36,320 --> 00:21:39,000 And for that, I'm going to turn it over to Ally, 505 00:21:39,000 --> 00:21:42,640 who is going to talk a little bit about what it's like to work at Uber. 506 00:21:42,640 --> 00:21:52,040 507 00:21:52,040 --> 00:21:53,570 ALLI: My name is Alli Reedy. 508 00:21:53,570 --> 00:21:55,580 I'm a university recruiter at Uber. 509 00:21:55,580 --> 00:22:01,520 I've been with Uber for a little bit, about 2.5 years now. 510 00:22:01,520 --> 00:22:06,620 Started out the university in general is fairly new at Uber. 511 00:22:06,620 --> 00:22:09,650 It's only been around for about 2.5 years. 512 00:22:09,650 --> 00:22:12,701 So we helped build this program from the ground up. 513 00:22:12,701 --> 00:22:15,200 So I'm going to tell you a little bit about Uber in general, 514 00:22:15,200 --> 00:22:17,960 some of the opportunities that we do have available. 515 00:22:17,960 --> 00:22:21,830 As I mentioned, we will have a Q&A at the end, 516 00:22:21,830 --> 00:22:24,440 as well as we'll stay after and answer any questions that you 517 00:22:24,440 --> 00:22:28,060 may have one-on-one. 518 00:22:28,060 --> 00:22:31,960 You can have a huge impact, obviously, if you join Uber. 519 00:22:31,960 --> 00:22:36,160 More people earn income from Uber than any other privately-held company 520 00:22:36,160 --> 00:22:41,390 other than McDonald's or Wal-Mart, which is pretty crazy. 521 00:22:41,390 --> 00:22:43,820 You can improve cities and communities. 522 00:22:43,820 --> 00:22:50,570 So obviously, we have seen fewer cars on the road, more decreases of congestion 523 00:22:50,570 --> 00:22:52,260 and things like that. 524 00:22:52,260 --> 00:22:55,910 It also frees up parking garages, real estate, 525 00:22:55,910 --> 00:22:59,680 and in business and residential space. 526 00:22:59,680 --> 00:23:01,430 I would say one thing that we're extremely 527 00:23:01,430 --> 00:23:04,550 proud of is that we are making cities safer. 528 00:23:04,550 --> 00:23:09,050 So 11% decrease in DUI-related car accidents 529 00:23:09,050 --> 00:23:11,570 after it launched in Seattle, which is pretty crazy. 530 00:23:11,570 --> 00:23:15,420 And I think we've been seeing it in a lot of cities in other cities as well. 531 00:23:15,420 --> 00:23:20,970 There's also been a 20% lower crime rate related just to transportation, 532 00:23:20,970 --> 00:23:27,320 such as your traditional taxis or things like that once we launched in Chicago. 533 00:23:27,320 --> 00:23:29,750 So same thing, we've seen that in a bunch of other cities 534 00:23:29,750 --> 00:23:31,770 as well, which is awesome. 535 00:23:31,770 --> 00:23:35,147 I think the cool part of that is if you are taking Uber, 536 00:23:35,147 --> 00:23:36,980 you know who you're getting in the car with. 537 00:23:36,980 --> 00:23:39,950 You know where your payment is going, and you know that the trip 538 00:23:39,950 --> 00:23:40,910 that you're taking. 539 00:23:40,910 --> 00:23:45,360 So it has reduced in that type of way as well. 540 00:23:45,360 --> 00:23:49,740 How many of you guys have heard of ATG, or Advanced Technology Group. 541 00:23:49,740 --> 00:23:50,540 Yeah? 542 00:23:50,540 --> 00:23:51,290 Awesome. 543 00:23:51,290 --> 00:23:54,470 So that is obviously new to Uber. 544 00:23:54,470 --> 00:23:58,780 We are investing in the future with self-driving technology. 545 00:23:58,780 --> 00:24:03,020 ATG, like I said, is also known as Advanced Technology Group. 546 00:24:03,020 --> 00:24:06,260 I would say most of our team sits in Pittsburgh, but a lot of them 547 00:24:06,260 --> 00:24:10,310 have recently moved to San Francisco to work on this type of technology, 548 00:24:10,310 --> 00:24:13,400 which is awesome. 549 00:24:13,400 --> 00:24:17,420 UberEATS-- how many have used UberEATS? 550 00:24:17,420 --> 00:24:18,140 Yeah? 551 00:24:18,140 --> 00:24:18,830 Awesome. 552 00:24:18,830 --> 00:24:20,020 I use it everyday. 553 00:24:20,020 --> 00:24:21,270 It's amazing. 554 00:24:21,270 --> 00:24:26,450 But we've expanded to 31 more college campuses 555 00:24:26,450 --> 00:24:30,580 since August, which has been crazy. 556 00:24:30,580 --> 00:24:34,580 It says with a 1 billion run rate in October 2016. 557 00:24:34,580 --> 00:24:40,600 And we do hope to get to about 10 billion within the next year or so. 558 00:24:40,600 --> 00:24:43,240 Some engineering locations that we do have-- 559 00:24:43,240 --> 00:24:46,440 our headquarters is obviously in San Francisco. 560 00:24:46,440 --> 00:24:51,670 We do have three offices there, actually all about 15 minutes walking distance 561 00:24:51,670 --> 00:24:53,260 from each other. 562 00:24:53,260 --> 00:24:56,020 We recently have opened a new office in Palo Alto, 563 00:24:56,020 --> 00:24:59,260 which has been around for about six to eight months now. 564 00:24:59,260 --> 00:25:04,480 We have 1,700 engineers and growing total in those locations. 565 00:25:04,480 --> 00:25:08,380 Seattle, New York, and Colorado have a little bit less. 566 00:25:08,380 --> 00:25:13,480 Seattle is actually one of our most growing offices in the United States. 567 00:25:13,480 --> 00:25:16,060 There are about 130 engineers now. 568 00:25:16,060 --> 00:25:19,450 But we'll have quite a few in the next few months or so. 569 00:25:19,450 --> 00:25:22,970 New York is a little bit smaller, same with Colorado. 570 00:25:22,970 --> 00:25:29,310 Our Colorado office is fairly tiny, and they mostly work on our mass teams. 571 00:25:29,310 --> 00:25:33,180 Here's just a general list of some of our engineering teams 572 00:25:33,180 --> 00:25:35,940 that we do have available at Uber-- 573 00:25:35,940 --> 00:25:40,140 business intelligence platform, consumer products, core infrastructure, 574 00:25:40,140 --> 00:25:45,390 maps, marketplace, and real time systems, security, and then 575 00:25:45,390 --> 00:25:47,010 Uber for business. 576 00:25:47,010 --> 00:25:51,240 Like I said, this is just a few general teams that we have at Uber. 577 00:25:51,240 --> 00:25:56,450 We do have a lot more within these larger groups. 578 00:25:56,450 --> 00:26:00,860 Obviously, you want to belong to a place. 579 00:26:00,860 --> 00:26:04,320 And you want to make sure that you are involved in other groups 580 00:26:04,320 --> 00:26:06,730 when you do join the real world. 581 00:26:06,730 --> 00:26:11,670 So here's just a few of our groups that we do have at Uber. 582 00:26:11,670 --> 00:26:16,440 If you were to do an internship or even be a full time employee, 583 00:26:16,440 --> 00:26:18,330 you can join these groups. 584 00:26:18,330 --> 00:26:21,510 If you were an intern, we do host ERG, which is also 585 00:26:21,510 --> 00:26:23,730 known as Employee Resource Groups. 586 00:26:23,730 --> 00:26:29,430 We do have networking nights with these groups, so Los Ubers, Woman of Uber, 587 00:26:29,430 --> 00:26:32,820 Parenting at Uber, Shalom, Uber Pride. 588 00:26:32,820 --> 00:26:38,440 This is a photo of our Uber Pride parade that was held in San Francisco. 589 00:26:38,440 --> 00:26:41,960 It's a very fun event that Uber does. 590 00:26:41,960 --> 00:26:43,270 Yeah, we have a lot more. 591 00:26:43,270 --> 00:26:46,530 This is just to name a few. 592 00:26:46,530 --> 00:26:49,760 I would say we definitely have a lot of great benefits and perks 593 00:26:49,760 --> 00:26:51,150 at Uber as well. 594 00:26:51,150 --> 00:26:56,850 So as a full-time employee, you will be receiving full benefits-- 595 00:26:56,850 --> 00:27:00,750 medical health, dental, vision, et cetera. 596 00:27:00,750 --> 00:27:05,640 We do have a free breakfast, lunch, and dinner Monday through Friday. 597 00:27:05,640 --> 00:27:09,990 I would say breakfast is more of a cold breakfast where we have like yogurt 598 00:27:09,990 --> 00:27:14,190 and egg, cereal, bagels, et cetera. 599 00:27:14,190 --> 00:27:15,960 Lunch and dinner are catered. 600 00:27:15,960 --> 00:27:18,510 There's different meals every day. 601 00:27:18,510 --> 00:27:22,800 You will be receiving a certain amount of monthly Uber credits 602 00:27:22,800 --> 00:27:25,030 per month, which is awesome. 603 00:27:25,030 --> 00:27:27,600 It's probably my most favorite perk at Uber. 604 00:27:27,600 --> 00:27:31,470 And I would say living in San Francisco, Uber is life. 605 00:27:31,470 --> 00:27:34,110 I don't think we'd be able to get over the big hills 606 00:27:34,110 --> 00:27:39,810 or walk anywhere in the city without Uber, so using these credits 607 00:27:39,810 --> 00:27:41,370 is awesome. 608 00:27:41,370 --> 00:27:44,940 If you do run out of your credits, you still receive a certain percentage 609 00:27:44,940 --> 00:27:46,830 each month as well. 610 00:27:46,830 --> 00:27:50,370 You will be receiving a monthly stipend for your phone bill. 611 00:27:50,370 --> 00:27:54,762 We do have unlimited paid time off, also known as vacation. 612 00:27:54,762 --> 00:27:56,220 I think that's an awesome perk too. 613 00:27:56,220 --> 00:27:59,850 I don't think you really realize like how valuable a day off 614 00:27:59,850 --> 00:28:02,130 is once you hit the real world. 615 00:28:02,130 --> 00:28:06,580 So as long as you have manager approval, you should be good to go. 616 00:28:06,580 --> 00:28:10,500 Paid time off during the holidays, I think we're about 11 or 12, 617 00:28:10,500 --> 00:28:14,940 I want to say, paid time off holidays throughout the year. 618 00:28:14,940 --> 00:28:20,850 We do a lot of volunteer events just in San Francisco, New York, Seattle. 619 00:28:20,850 --> 00:28:22,297 Austin I know has done a few. 620 00:28:22,297 --> 00:28:23,130 Atlanta has as well. 621 00:28:23,130 --> 00:28:27,240 So that's some of our other offices. 622 00:28:27,240 --> 00:28:31,870 If you are interested in applying, you can go to this link. 623 00:28:31,870 --> 00:28:35,940 This link will basically show you some of the roles that we have available. 624 00:28:35,940 --> 00:28:40,080 If you're interested in a specific role, and it's maybe not posted on that page, 625 00:28:40,080 --> 00:28:41,790 feel free to come talk to me afterwards. 626 00:28:41,790 --> 00:28:46,584 I'm more than happy to talk with anybody about the interview process, what 627 00:28:46,584 --> 00:28:48,000 we look for, and things like that. 628 00:28:48,000 --> 00:28:50,610 629 00:28:50,610 --> 00:28:55,330 So we can do a quick Q&A. And then we will stay around for about 20, 630 00:28:55,330 --> 00:28:58,560 25 minutes afterwards if you have any questions. 631 00:28:58,560 --> 00:29:00,332 632 00:29:00,332 --> 00:29:06,990 YUKI: Are there any questions about either Uber, some of the material, 633 00:29:06,990 --> 00:29:07,490 whatever. 634 00:29:07,490 --> 00:29:09,150 We're down to hear it. 635 00:29:09,150 --> 00:29:11,419 Ask us anything, I guess. 636 00:29:11,419 --> 00:29:12,460 MATTHEW: Tech in general. 637 00:29:12,460 --> 00:29:14,430 638 00:29:14,430 --> 00:29:16,930 AUDIENCE: So this might not really be what you guys covered, 639 00:29:16,930 --> 00:29:19,750 but I was doing research on Uber and you were doing something 640 00:29:19,750 --> 00:29:23,956 about electric flying cars in terms of Uber Elevate. 641 00:29:23,956 --> 00:29:28,290 And I was wondering how can we get more involved in that 642 00:29:28,290 --> 00:29:31,660 if we wanted to work with Uber? 643 00:29:31,660 --> 00:29:36,850 YUKI: So the question, for those of you online, was about our flying cars 644 00:29:36,850 --> 00:29:41,440 initiative, Uber Elevate, and how to get more involved. 645 00:29:41,440 --> 00:29:43,160 It's really still a nascent effort. 646 00:29:43,160 --> 00:29:48,220 And it's just being part of the narrative of how we move people around. 647 00:29:48,220 --> 00:29:50,920 And that is definitely a future possibility. 648 00:29:50,920 --> 00:29:55,300 And so it's for that reason that we're involved in those kinds of efforts. 649 00:29:55,300 --> 00:29:57,610 At the moment, it's really just trying to figure out 650 00:29:57,610 --> 00:30:02,020 who the players are in the world who are in this space, talking to them, 651 00:30:02,020 --> 00:30:03,850 and figuring out what model is right. 652 00:30:03,850 --> 00:30:07,240 So it's just super small right now, and if you come to Uber, 653 00:30:07,240 --> 00:30:11,380 there's certainly opportunities to work with that team, 654 00:30:11,380 --> 00:30:13,040 to get to know that team. 655 00:30:13,040 --> 00:30:15,340 And I think that's the best way to do it. 656 00:30:15,340 --> 00:30:19,401 But I think we'll find out more as we learn more about the technology 657 00:30:19,401 --> 00:30:19,900 honestly. 658 00:30:19,900 --> 00:30:23,882 659 00:30:23,882 --> 00:30:28,280 AUDIENCE: You guys talked about from when someone requests a ride 660 00:30:28,280 --> 00:30:29,490 to whey they get in the car. 661 00:30:29,490 --> 00:30:31,610 There's also this other step of matching. 662 00:30:31,610 --> 00:30:36,233 There's this two-sided matching problem that you have to solve. 663 00:30:36,233 --> 00:30:38,787 Can you maybe talk about features in figuring it out. 664 00:30:38,787 --> 00:30:41,495 Are there any interesting features you guys used in that respect, 665 00:30:41,495 --> 00:30:45,670 besides distance from driver? 666 00:30:45,670 --> 00:30:49,170 YUKI: The question was about matching and what 667 00:30:49,170 --> 00:30:54,200 are the things that we do to intelligently match riders and drivers, 668 00:30:54,200 --> 00:30:57,600 as well as riders and riders in the case of uberPOOL. 669 00:30:57,600 --> 00:31:00,600 I worked a little bit on uberPOOL, so this can speak to it. 670 00:31:00,600 --> 00:31:05,500 I think fundamentally, it's one big optimization problem. 671 00:31:05,500 --> 00:31:07,270 And in the case of uberPOOL, for example, 672 00:31:07,270 --> 00:31:12,810 we're trying to find riders whose journey overlaps most with you. 673 00:31:12,810 --> 00:31:15,465 And that's what economically makes sense for us. 674 00:31:15,465 --> 00:31:18,090 Because the greater the overlap, that means that we're actually 675 00:31:18,090 --> 00:31:19,048 being really efficient. 676 00:31:19,048 --> 00:31:21,090 And so the objective function there effectively 677 00:31:21,090 --> 00:31:25,920 is keeping all car seats populated at any given time. 678 00:31:25,920 --> 00:31:30,480 And so that's what we're striving for. 679 00:31:30,480 --> 00:31:32,610 And there are some interesting nuances around. 680 00:31:32,610 --> 00:31:37,690 For example, how long if we increase the window for matching, as an example, 681 00:31:37,690 --> 00:31:42,270 then it increases their likelihood of being able to produce a better match. 682 00:31:42,270 --> 00:31:46,510 But that might mean that there's a match that's comes in and that's available, 683 00:31:46,510 --> 00:31:50,350 but you might predict that you'll in 10 seconds, come up with a better match 684 00:31:50,350 --> 00:31:52,760 that you might just discard it and wait for the next one. 685 00:31:52,760 --> 00:31:54,630 So it's this game of probability. 686 00:31:54,630 --> 00:31:57,170 687 00:31:57,170 --> 00:32:00,870 And those are the experiments that we do on the rider and driver side as well. 688 00:32:00,870 --> 00:32:04,251 Sometimes we tell you in the app today that, OK, a car is on the way, 689 00:32:04,251 --> 00:32:06,750 and you have a two minute ETA, but you're not assigned a car 690 00:32:06,750 --> 00:32:08,080 until a little bit later. 691 00:32:08,080 --> 00:32:09,690 And that's to give us a little bit of breathing room 692 00:32:09,690 --> 00:32:10,981 to find the most optimal thing. 693 00:32:10,981 --> 00:32:13,324 So there are some tricks like that we use, 694 00:32:13,324 --> 00:32:15,990 and certainly many, many other factors that go into figuring out 695 00:32:15,990 --> 00:32:18,248 what the best optimal route is. 696 00:32:18,248 --> 00:32:20,846 MATTHEW: I think the other major component to how we're going 697 00:32:20,846 --> 00:32:23,721 to actually perform that match-- and there's tons of features that go 698 00:32:23,721 --> 00:32:25,190 into making that decision-- 699 00:32:25,190 --> 00:32:26,440 one of the major ones is time. 700 00:32:26,440 --> 00:32:28,320 So we look and see what is the time it's going to take 701 00:32:28,320 --> 00:32:29,730 for the car to get to a given rider. 702 00:32:29,730 --> 00:32:32,896 Because we want to minimize that time as much as we can for all the pickups, 703 00:32:32,896 --> 00:32:36,480 so your pickup quality increases, and your wait time is reduced. 704 00:32:36,480 --> 00:32:39,570 We also want to minimize the time the drivers spend going to pick you up. 705 00:32:39,570 --> 00:32:42,150 Because we want the driver to have as much of their time 706 00:32:42,150 --> 00:32:44,929 as they are online at Uber, to have their car be full of drivers, 707 00:32:44,929 --> 00:32:47,970 with passengers, that they're making money going from point A to point B. 708 00:32:47,970 --> 00:32:50,610 And so if they're spending more time driving to a pick up, 709 00:32:50,610 --> 00:32:52,240 they're spending less time earning. 710 00:32:52,240 --> 00:32:57,510 So we're also trying to optimize around making most [INAUDIBLE] of that time. 711 00:32:57,510 --> 00:33:00,720 And then there's a whole other component on top of that time. 712 00:33:00,720 --> 00:33:03,390 There's a lot of complexity with the pricing of the marketplace 713 00:33:03,390 --> 00:33:05,707 itself, which is more outside of both of our domains. 714 00:33:05,707 --> 00:33:08,790 There's an entire team of just folks that just takes these various inputs, 715 00:33:08,790 --> 00:33:15,660 and then performs an optimization, basically the drivers time. 716 00:33:15,660 --> 00:33:19,382 And also the amount that we're paying the driver, and the estimated 717 00:33:19,382 --> 00:33:21,090 fare that we're receiving from the rider. 718 00:33:21,090 --> 00:33:23,430 So those are the components that are at play. 719 00:33:23,430 --> 00:33:27,310 And it's constantly changing how the optimization is being performed. 720 00:33:27,310 --> 00:33:29,210 And it varies also by product. 721 00:33:29,210 --> 00:33:30,995 That's a really good question though. 722 00:33:30,995 --> 00:33:32,515 AUDIENCE: I was wondering if you guys could talk a little bit more 723 00:33:32,515 --> 00:33:35,160 personally to your roles as product managers, 724 00:33:35,160 --> 00:33:37,372 and maybe how that works on a day in, day out basis. 725 00:33:37,372 --> 00:33:39,830 Perhaps you could compare it to a software engineering role 726 00:33:39,830 --> 00:33:41,642 if you guys have any experience with that. 727 00:33:41,642 --> 00:33:46,360 728 00:33:46,360 --> 00:33:48,564 MATTHEW: The question was can we describe 729 00:33:48,564 --> 00:33:50,730 a bit what it's like to work with a product manager, 730 00:33:50,730 --> 00:33:53,800 maybe compare it with like an IC role, like a software engineer. 731 00:33:53,800 --> 00:33:56,920 So my experience at Uber and actually at other companies as well. 732 00:33:56,920 --> 00:33:58,540 I worked at Google before. 733 00:33:58,540 --> 00:34:01,310 A typical setup for a team structure is you're 734 00:34:01,310 --> 00:34:05,290 going have a product manager, software engineers, data science, 735 00:34:05,290 --> 00:34:09,130 and, if you're working on a product that has some sort of experience component, 736 00:34:09,130 --> 00:34:13,210 a design designer and maybe a user researcher. 737 00:34:13,210 --> 00:34:16,239 The project manager in this sense is the glue of this group. 738 00:34:16,239 --> 00:34:19,780 You're working to leverage the skills that each person brings to the table, 739 00:34:19,780 --> 00:34:24,770 and try to effectively move toward some shared vision utilizing those skills. 740 00:34:24,770 --> 00:34:27,190 So there's a real tactical component, just day-to-day, 741 00:34:27,190 --> 00:34:29,440 making sure that those folks are all working together, 742 00:34:29,440 --> 00:34:31,590 thing are being unblocked and they're on time. 743 00:34:31,590 --> 00:34:33,460 They're kind of like project manage-y. 744 00:34:33,460 --> 00:34:35,020 But then at a high level, you're also helping 745 00:34:35,020 --> 00:34:37,561 to develop what's the thing that we all should be working on. 746 00:34:37,561 --> 00:34:39,310 What's that vision for this group. 747 00:34:39,310 --> 00:34:41,170 And that's really where you're taking input 748 00:34:41,170 --> 00:34:45,489 both from your team, and from outside of industry, from research, 749 00:34:45,489 --> 00:34:49,540 to help [INAUDIBLE] what you're doing long term. 750 00:34:49,540 --> 00:34:50,750 You have to wear both hats. 751 00:34:50,750 --> 00:34:53,710 And the best product managers do both of those things. 752 00:34:53,710 --> 00:34:54,790 But it's fun, because you get to be in the weeds, 753 00:34:54,790 --> 00:34:56,496 but you also get to be kind of high level. 754 00:34:56,496 --> 00:34:57,310 That's my take on it. 755 00:34:57,310 --> 00:34:59,726 YUKI: Yeah, the way I think about some of the distinctions 756 00:34:59,726 --> 00:35:02,890 is, ultimately, a product manager is responsible for the why. 757 00:35:02,890 --> 00:35:03,980 Why are we doing this? 758 00:35:03,980 --> 00:35:05,860 Why should we invest in this? 759 00:35:05,860 --> 00:35:08,890 The what comes from many parts of the company. 760 00:35:08,890 --> 00:35:11,000 There are ideas floating around everywhere. 761 00:35:11,000 --> 00:35:14,230 And in terms of the how, that's one of the places where the engineers are 762 00:35:14,230 --> 00:35:17,380 really involved in and designers in figuring out exactly how we're solving 763 00:35:17,380 --> 00:35:19,426 this particular solution or problem. 764 00:35:19,426 --> 00:35:21,800 But it's the product manager to decide what problem we're 765 00:35:21,800 --> 00:35:23,720 solving in the first place. 766 00:35:23,720 --> 00:35:24,950 So that's one thing. 767 00:35:24,950 --> 00:35:29,500 And even in a pickup space, a bad pick up is a problem. 768 00:35:29,500 --> 00:35:32,816 But maybe our jobs might be to break that down into specific areas, 769 00:35:32,816 --> 00:35:34,690 like what part of that pickup is the problem. 770 00:35:34,690 --> 00:35:38,510 And then apply our team to figure out solutions for it. 771 00:35:38,510 --> 00:35:40,524 So that's one way I think about it. 772 00:35:40,524 --> 00:35:45,300 MATTHEW: And comparing it to an IC role, I did work in an IC role at Google. 773 00:35:45,300 --> 00:35:50,560 Sorry, individual contributor, working basically as a component to that team, 774 00:35:50,560 --> 00:35:53,390 as opposed to being a manager. 775 00:35:53,390 --> 00:35:56,917 Some of the differences are when you are an IC, you problem space is 776 00:35:56,917 --> 00:35:58,000 a little bit more defined. 777 00:35:58,000 --> 00:36:00,030 You spend more of your time on your own, not 778 00:36:00,030 --> 00:36:02,650 in meetings, just working on that problem, 779 00:36:02,650 --> 00:36:06,520 writing code, performing analyses, or whatever you might be doing. 780 00:36:06,520 --> 00:36:10,540 So it's a lot more like you're spending time with fewer people 781 00:36:10,540 --> 00:36:13,150 and more with an idea and tools. 782 00:36:13,150 --> 00:36:16,300 Whereas when you're a PM, we spend a lot more of our time with people, 783 00:36:16,300 --> 00:36:22,370 and talking, and writing documents, and clarifying, 784 00:36:22,370 --> 00:36:23,950 roadmapping why you're doing things. 785 00:36:23,950 --> 00:36:26,908 So you spend more time with ideas and people, as opposed to with tools. 786 00:36:26,908 --> 00:36:29,170 So it's just a different way of spending time. 787 00:36:29,170 --> 00:36:30,160 They're bot satisfying. 788 00:36:30,160 --> 00:36:36,759 And the skills you if you work in an IC capacity as an engineer, 789 00:36:36,759 --> 00:36:40,050 those skills are totally transferable to management and backwards and forwards. 790 00:36:40,050 --> 00:36:42,508 YUKI: And to answer your second part of the question of how 791 00:36:42,508 --> 00:36:46,300 it's different from a place like Google, or I also worked at Microsoft. 792 00:36:46,300 --> 00:36:51,200 I would say the biggest thing is the operational nature of Uber's work. 793 00:36:51,200 --> 00:36:55,300 So for example, we experimented a couple of years ago-- 794 00:36:55,300 --> 00:36:58,090 and this is still part of the experience today-- 795 00:36:58,090 --> 00:37:02,080 in New York, getting people to walk to the street corners. 796 00:37:02,080 --> 00:37:05,590 And that would be more efficient, so cars can go up and down avenues, 797 00:37:05,590 --> 00:37:08,170 instead of driving all over the place. 798 00:37:08,170 --> 00:37:10,270 And so that's the kind of thing where we need 799 00:37:10,270 --> 00:37:12,430 to work with the local team on the ground 800 00:37:12,430 --> 00:37:15,014 to make sure that everyone understands what's happening. 801 00:37:15,014 --> 00:37:16,930 They might have to change their support flows, 802 00:37:16,930 --> 00:37:19,330 or they might be getting a lot of inbound about it. 803 00:37:19,330 --> 00:37:21,880 And so there's aspect of this physical world 804 00:37:21,880 --> 00:37:25,490 and the team on the ground that you have to be really thinking about, 805 00:37:25,490 --> 00:37:27,160 which is very different at Uber. 806 00:37:27,160 --> 00:37:30,680 I felt that at Microsoft or Google, you could do this all remotely, 807 00:37:30,680 --> 00:37:32,320 and manage it, and it's fine. 808 00:37:32,320 --> 00:37:34,130 So that's one big difference. 809 00:37:34,130 --> 00:37:37,030 And another one is just the size. 810 00:37:37,030 --> 00:37:41,110 Uber is a big company now, but even then, it still feels like a startup. 811 00:37:41,110 --> 00:37:44,650 And one of the things about that is, as a product manager, 812 00:37:44,650 --> 00:37:48,954 you're responsible for your team, and how well they're doing, 813 00:37:48,954 --> 00:37:52,120 and how they're feeling, and all these other aspects of even like recruiting 814 00:37:52,120 --> 00:37:54,280 and things like that. 815 00:37:54,280 --> 00:37:56,090 you're given a much larger responsibility. 816 00:37:56,090 --> 00:37:59,338 So that's something I felt as I've transitioned from different companies. 817 00:37:59,338 --> 00:38:03,820 818 00:38:03,820 --> 00:38:05,812 AUDIENCE: So over the last few years, I've just 819 00:38:05,812 --> 00:38:08,510 been transitioning from a scrappy startup that's 820 00:38:08,510 --> 00:38:11,710 disrupting a massive space to a more established company, 821 00:38:11,710 --> 00:38:14,440 kind of as you were touching on briefly. 822 00:38:14,440 --> 00:38:18,160 I'm curious how things are changing at Uber, like structurally, 823 00:38:18,160 --> 00:38:23,000 like relationship, how it flows between engineers and management. 824 00:38:23,000 --> 00:38:27,150 What the individual's role looks like now for someone like me 825 00:38:27,150 --> 00:38:30,775 who is trying to find a position right out of college. 826 00:38:30,775 --> 00:38:32,650 How has that changed over the last few years? 827 00:38:32,650 --> 00:38:35,300 How is it projected to change? 828 00:38:35,300 --> 00:38:37,560 YUKI: So the question was about Uber as a company, 829 00:38:37,560 --> 00:38:41,450 it's becoming bigger and growing, and how the dynamics of the workplace 830 00:38:41,450 --> 00:38:43,200 changed as a result of that. and what it's 831 00:38:43,200 --> 00:38:45,180 like as am individual contributor working 832 00:38:45,180 --> 00:38:46,720 within that kind of environment. 833 00:38:46,720 --> 00:38:49,720 So I can take that part of it. 834 00:38:49,720 --> 00:38:52,700 So I think, yeah, it's definitely true. 835 00:38:52,700 --> 00:38:56,640 When Matt and I joined, we were about 2000 people, and now we're like 15,000. 836 00:38:56,640 --> 00:38:59,100 And so that definitely changes a lot of things. 837 00:38:59,100 --> 00:39:05,100 And it's certainly the case that we have real-world responsibilities now. 838 00:39:05,100 --> 00:39:09,210 We can't just launch something in some random city and not tell anyone. 839 00:39:09,210 --> 00:39:12,229 And that kind of thing happened a lot when there 840 00:39:12,229 --> 00:39:13,770 were a lot of these autonomous teams. 841 00:39:13,770 --> 00:39:18,360 So there is a certain aspect of we do need to be talking with each other 842 00:39:18,360 --> 00:39:19,140 and communicating. 843 00:39:19,140 --> 00:39:24,060 And there certainly needs to be some process in place, even a marketing 844 00:39:24,060 --> 00:39:27,000 campaign that one city runs that actually people 845 00:39:27,000 --> 00:39:29,375 need to know because they could affect many other people. 846 00:39:29,375 --> 00:39:31,916 You could have competing policies, all these kinds of things. 847 00:39:31,916 --> 00:39:33,930 So there's the reality that as we grow bigger 848 00:39:33,930 --> 00:39:36,612 and we have millions of millions of riders and millions 849 00:39:36,612 --> 00:39:38,820 and millions of drivers whose lives we're supporting, 850 00:39:38,820 --> 00:39:40,750 we have to put some process in place. 851 00:39:40,750 --> 00:39:42,750 So for that, I think there's a little bit more 852 00:39:42,750 --> 00:39:45,060 of the checks and balances that are necessary. 853 00:39:45,060 --> 00:39:47,820 But I think parts of us are growing that we're 854 00:39:47,820 --> 00:39:51,090 really conscious of is making sure that everyone still 855 00:39:51,090 --> 00:39:53,050 feels that sense of autonomy. 856 00:39:53,050 --> 00:39:55,290 And I think the way we're organized is actually 857 00:39:55,290 --> 00:39:58,129 a good way that facilitates that, which is if you're in a team, 858 00:39:58,129 --> 00:40:00,420 let's say there's a team called the pick up experience. 859 00:40:00,420 --> 00:40:03,480 And their sole mission is to make the pickup great. 860 00:40:03,480 --> 00:40:06,420 And so they actually have a full stock set 861 00:40:06,420 --> 00:40:12,660 of engineers across both Android and iOS, as well as deep inside the mapping 862 00:40:12,660 --> 00:40:13,740 stack. 863 00:40:13,740 --> 00:40:18,780 There is a person an ops, product manager, designers, data scientists. 864 00:40:18,780 --> 00:40:21,360 And they basically have the expertise to make 865 00:40:21,360 --> 00:40:23,810 this happen without creating too many more dependencies. 866 00:40:23,810 --> 00:40:27,510 So that's how Uber is organized in terms of teams 867 00:40:27,510 --> 00:40:32,890 around these missions that can execute on their own. 868 00:40:32,890 --> 00:40:35,609 And so the setup has this autonomy. 869 00:40:35,609 --> 00:40:37,650 And it's really then just a matter of making sure 870 00:40:37,650 --> 00:40:40,191 that there is a process around that in terms of communication 871 00:40:40,191 --> 00:40:42,720 where we're all talking, so it's not always silent. 872 00:40:42,720 --> 00:40:44,802 So it's definitely a delicate balancing act. 873 00:40:44,802 --> 00:40:47,760 And I would say that one of the interesting things right now of joining 874 00:40:47,760 --> 00:40:51,660 is that you're in this kind of like gray space 875 00:40:51,660 --> 00:40:56,220 where we're trying to figure out what the most effective way to do this is 876 00:40:56,220 --> 00:40:59,400 and contributing to that process. 877 00:40:59,400 --> 00:41:03,060 For example, at a place like Google, they have everything figured out. 878 00:41:03,060 --> 00:41:10,200 So there's this launch process that's quite sophisticated. 879 00:41:10,200 --> 00:41:12,300 But maybe that's not the right approach for Uber, 880 00:41:12,300 --> 00:41:13,450 and we're still trying to figure out. 881 00:41:13,450 --> 00:41:16,080 And that's part of what it means to come and work at Uber. 882 00:41:16,080 --> 00:41:18,750 MATTHEW: And one other thing to consider too, especially as a recent grad, 883 00:41:18,750 --> 00:41:20,624 is the kind of mentorship you're going to get 884 00:41:20,624 --> 00:41:23,860 and experiences you're going to have when you're taking on that first job. 885 00:41:23,860 --> 00:41:27,660 And I think one of the benefits of Uber growing over the past few years, 886 00:41:27,660 --> 00:41:30,510 and now having doubled and tripled in size since we've joined, 887 00:41:30,510 --> 00:41:35,790 is that there is more people to actually mentor and help people grow. 888 00:41:35,790 --> 00:41:39,625 In the earlier days, you showed up, and it was like, all right, 889 00:41:39,625 --> 00:41:41,250 figure out how to make this thing work. 890 00:41:41,250 --> 00:41:43,583 And there's no one who's really going to hold your hand. 891 00:41:43,583 --> 00:41:46,465 And it was a lot to figuring it out on your own, which is find. 892 00:41:46,465 --> 00:41:47,340 That's a way to grow. 893 00:41:47,340 --> 00:41:50,332 But from my experience, having come from Google, 894 00:41:50,332 --> 00:41:52,040 it was really nice to be in a place where 895 00:41:52,040 --> 00:41:55,410 there was a layer of management who were people who were a few years older 896 00:41:55,410 --> 00:41:57,690 than me who would teach me how to do x, y and z, 897 00:41:57,690 --> 00:41:59,074 and help me hone my own skills. 898 00:41:59,074 --> 00:42:01,740 I think as we're maturing as a company, we're getting to a place 899 00:42:01,740 --> 00:42:02,865 where we have those people. 900 00:42:02,865 --> 00:42:06,600 People have time to be able to take on recent grads and help train them 901 00:42:06,600 --> 00:42:08,290 and onboard them an stuff. 902 00:42:08,290 --> 00:42:10,700 And that is a really important part of any first job, 903 00:42:10,700 --> 00:42:12,010 just having that mentorship. 904 00:42:12,010 --> 00:42:13,110 And I think that's something that we now are 905 00:42:13,110 --> 00:42:14,610 able to offer, which is really cool. 906 00:42:14,610 --> 00:42:20,110 907 00:42:20,110 --> 00:42:22,430 AUDIENCE: How does internships work at Uber? 908 00:42:22,430 --> 00:42:25,184 And what do you look for in an intern? 909 00:42:25,184 --> 00:42:32,330 910 00:42:32,330 --> 00:42:36,800 ALLI: The question was, what is an internship program like, 911 00:42:36,800 --> 00:42:38,520 and what do we look for in an intern? 912 00:42:38,520 --> 00:42:39,019 Correct? 913 00:42:39,019 --> 00:42:40,220 AUDIENCE: Yeah. 914 00:42:40,220 --> 00:42:43,840 ALLI: I'll tell you a little bit about our internship program in general. 915 00:42:43,840 --> 00:42:47,660 So typically, it's about a 12 to 16-week internship, 916 00:42:47,660 --> 00:42:50,390 depending on which school you go to. 917 00:42:50,390 --> 00:42:53,260 We do hire interns year round. 918 00:42:53,260 --> 00:42:56,550 I would say our biggest intern class is obviously in the summer. 919 00:42:56,550 --> 00:42:59,120 But then we do have a fall intern class. 920 00:42:59,120 --> 00:43:03,590 I believe our fall intern class is about 35 interns right now. 921 00:43:03,590 --> 00:43:08,030 And then in the spring, it's roughly about the same too. 922 00:43:08,030 --> 00:43:12,980 Over summer, we probably have about 180 and growing. 923 00:43:12,980 --> 00:43:17,724 So there's obviously a lot more intern events and things 924 00:43:17,724 --> 00:43:18,890 like that during the summer. 925 00:43:18,890 --> 00:43:21,260 So that's the program itself. 926 00:43:21,260 --> 00:43:24,200 927 00:43:24,200 --> 00:43:27,710 There's a few different internships that we have available. 928 00:43:27,710 --> 00:43:32,300 So if you are looking for a software engineer internship, 929 00:43:32,300 --> 00:43:36,590 basically what will happen is you will interview as a generalist. 930 00:43:36,590 --> 00:43:38,540 And what will happen is that we will allocate 931 00:43:38,540 --> 00:43:43,520 you a team, most likely about three months prior to starting. 932 00:43:43,520 --> 00:43:51,050 And the reason we do that is just because Uber is constantly changing, 933 00:43:51,050 --> 00:43:53,970 organizing team, things like that. 934 00:43:53,970 --> 00:43:57,920 So we'll allocate you to a team three months prior. 935 00:43:57,920 --> 00:44:00,480 You actually get to choose your team too. 936 00:44:00,480 --> 00:44:05,010 We typically give you a survey of about 20 to 25 that you can choose from. 937 00:44:05,010 --> 00:44:09,054 I would say most interns did get their top two this year. 938 00:44:09,054 --> 00:44:10,970 If they didn't get their top two, it was maybe 939 00:44:10,970 --> 00:44:14,540 because like they wanted UberEATS, and they wanted Palo Alto. 940 00:44:14,540 --> 00:44:18,000 UberEATS isn't in Palo Alto, so they had to work in San Francisco. 941 00:44:18,000 --> 00:44:19,820 So things like that. 942 00:44:19,820 --> 00:44:24,646 Everybody loves having an intern, which is awesome. 943 00:44:24,646 --> 00:44:25,520 Yep, you've had them. 944 00:44:25,520 --> 00:44:27,260 YUKI: Yea, I've had them. 945 00:44:27,260 --> 00:44:29,540 I've had the the maps routing team. 946 00:44:29,540 --> 00:44:35,720 And we have had the most phenomenal interns every quarter, every semester. 947 00:44:35,720 --> 00:44:36,620 You'd be amazed. 948 00:44:36,620 --> 00:44:39,453 So I used to work at Google, so I've had interns at other companies. 949 00:44:39,453 --> 00:44:42,140 You'd be amazed the things at Uber that the interns build. 950 00:44:42,140 --> 00:44:45,130 Things that you've all used in the app, interns have totally built 951 00:44:45,130 --> 00:44:48,129 and they've launched within their internships, which just blows my mind. 952 00:44:48,129 --> 00:44:51,240 Because other bigger companies, it's just hard to ramp up in three months 953 00:44:51,240 --> 00:44:52,406 and actually ship something. 954 00:44:52,406 --> 00:44:56,180 But we have interns contribute really amazing features to the app. 955 00:44:56,180 --> 00:44:58,557 So I've been blown away with some of our interns. 956 00:44:58,557 --> 00:44:59,890 Most of them have come back now. 957 00:44:59,890 --> 00:45:00,620 Some of them are still in school. 958 00:45:00,620 --> 00:45:01,670 But yeah, pretty cool. 959 00:45:01,670 --> 00:45:04,820 ALLI: That's one of the things I always tell our interns as well. 960 00:45:04,820 --> 00:45:07,190 If you're looking to make an impact at a company, 961 00:45:07,190 --> 00:45:09,830 and you're looking to work on meaningful work, 962 00:45:09,830 --> 00:45:13,940 and you're looking to work on a project that really interests you, 963 00:45:13,940 --> 00:45:16,280 then Uber is probably the place to be. 964 00:45:16,280 --> 00:45:18,610 If you're looking for an internship that's like, 965 00:45:18,610 --> 00:45:23,390 I don't really feel like working today, or I want fun all day 966 00:45:23,390 --> 00:45:26,730 long, that probably isn't the place for you. 967 00:45:26,730 --> 00:45:29,720 And we've done a pretty good job at making sure 968 00:45:29,720 --> 00:45:32,740 that each intern works on different projects that impact 969 00:45:32,740 --> 00:45:35,030 the company like Matt was just saying. 970 00:45:35,030 --> 00:45:37,850 We've had a few interns work on projects that have been 971 00:45:37,850 --> 00:45:40,580 implemented into the entire company. 972 00:45:40,580 --> 00:45:43,970 I think we've have an intern speak at our All Hands meetings 973 00:45:43,970 --> 00:45:49,320 about the type of project that they worked on within a 12-week span. 974 00:45:49,320 --> 00:45:53,960 So I wouldn't say that we look for any specific things. 975 00:45:53,960 --> 00:45:57,440 I would say we really are looking for passion 976 00:45:57,440 --> 00:46:00,590 and looking for somebody who wants to learn 977 00:46:00,590 --> 00:46:02,630 and who wants to grow their career. 978 00:46:02,630 --> 00:46:07,340 From most of my interns that I've had over the past 2.5 years, 979 00:46:07,340 --> 00:46:10,760 I would say majority of them have said this is the most challenging internship 980 00:46:10,760 --> 00:46:11,600 they've had. 981 00:46:11,600 --> 00:46:14,600 Because they have never learned so much within a 12-week span. 982 00:46:14,600 --> 00:46:16,920 983 00:46:16,920 --> 00:46:19,420 MATTHEW: You're basically treated like a full-time employee. 984 00:46:19,420 --> 00:46:21,451 It's pretty cool. 985 00:46:21,451 --> 00:46:23,450 AUDIENCE: I'm asking this for my little brother. 986 00:46:23,450 --> 00:46:26,477 Do you guys do freshman interns after freshman year? 987 00:46:26,477 --> 00:46:27,810 ALLI: We do-- freshman friendly. 988 00:46:27,810 --> 00:46:30,227 989 00:46:30,227 --> 00:46:32,143 AUDIENCE: Do you guys draw a clear distinction 990 00:46:32,143 --> 00:46:34,710 between project managers and individual contributors, 991 00:46:34,710 --> 00:46:38,267 or is there an opportunity to do both? 992 00:46:38,267 --> 00:46:41,350 YUKI: The question was about product managers and individual contributors. 993 00:46:41,350 --> 00:46:45,250 I think individual contributors in that context is a little bit of a misnomer. 994 00:46:45,250 --> 00:46:48,640 Product managers are individual contributors, too, in some ways. 995 00:46:48,640 --> 00:46:53,200 It just so happens that you're working with many people. 996 00:46:53,200 --> 00:46:55,390 And so that's the primary distinction. 997 00:46:55,390 --> 00:46:58,240 But there are definitely managers of engineers, 998 00:46:58,240 --> 00:47:02,710 managers of product managers, and things like that, especially off the bat. 999 00:47:02,710 --> 00:47:05,262 Everyone is essentially an individual contributor. 1000 00:47:05,262 --> 00:47:08,160 1001 00:47:08,160 --> 00:47:10,760 All right, we'll stick around. 1002 00:47:10,760 --> 00:47:13,180 But thanks for showing up. 1003 00:47:13,180 --> 00:47:16,200