1 00:00:00,000 --> 00:00:02,000 Hi and welcome to the Neil Ashton 2 00:00:02,000 --> 00:00:04,000 podcast. In each episode, we explain 3 00:00:04,000 --> 00:00:05,000 some of the fascinating ways that 4 00:00:05,000 --> 00:00:07,000 [music] science and engineering are 5 00:00:07,000 --> 00:00:10,000 changing the world around us. We talk to 6 00:00:10,000 --> 00:00:12,000 leading engineers from elite level 7 00:00:12,000 --> 00:00:15,000 sports like cycling and Formula 1 to 8 00:00:15,000 --> 00:00:17,000 some of the world's top academics to 9 00:00:17,000 --> 00:00:19,000 understand how fluid dynamics, machine 10 00:00:19,000 --> 00:00:21,000 learning, supercomputing are bringing 11 00:00:21,000 --> 00:00:24,000 [music] in a new era of discovery. We 12 00:00:24,000 --> 00:00:26,000 also hear some of their life stories, 13 00:00:26,000 --> 00:00:28,000 their career advice, and lessons they've 14 00:00:28,000 --> 00:00:30,000 learned on the way that I hope will be 15 00:00:30,000 --> 00:00:33,000 helpful to you, too. So, sit back and 16 00:00:33,000 --> 00:00:39,000 enjoy this episode. 17 00:00:39,000 --> 00:00:41,000 Hi, welcome back to the Neil Ashton 18 00:00:41,000 --> 00:00:44,000 podcast. So, as the title of this uh 19 00:00:44,000 --> 00:00:46,000 episode suggests, I've changed jobs. 20 00:00:46,000 --> 00:00:49,000 I've left Nvidia and I'm joining uh 21 00:00:49,000 --> 00:00:54,000 Mistral. I wanted to use this episode to 22 00:00:54,000 --> 00:00:55,000 talk a little bit more about the reasons 23 00:00:55,000 --> 00:00:59,000 for doing it but but more than that go 24 00:00:59,000 --> 00:01:02,000 into talking a little bit more from a 25 00:01:02,000 --> 00:01:04,000 career progression point of view uh 26 00:01:04,000 --> 00:01:08,000 career advice point of view um I've done 27 00:01:08,000 --> 00:01:09,000 episodes in the past you know which have 28 00:01:09,000 --> 00:01:11,000 tried to give some guidance to people 29 00:01:11,000 --> 00:01:13,000 who are maybe early on in their careers 30 00:01:13,000 --> 00:01:15,000 uh just after university or first jobs 31 00:01:15,000 --> 00:01:17,000 uh and I thought it might be useful a 32 00:01:17,000 --> 00:01:20,000 time to reflect and and maybe give some 33 00:01:20,000 --> 00:01:22,000 things I've learned uh along the way and 34 00:01:22,000 --> 00:01:24,000 to briefly talk about some of the sort 35 00:01:24,000 --> 00:01:28,000 of reasons for me doing it. Um this is 36 00:01:28,000 --> 00:01:30,000 in no way a sort of commercial thing. 37 00:01:30,000 --> 00:01:32,000 I'm not trying to sell you to to to 38 00:01:32,000 --> 00:01:34,000 Mistral. Um I just in the in 39 00:01:34,000 --> 00:01:36,000 transparency I I think this actually 40 00:01:36,000 --> 00:01:39,000 might be useful for some people to know. 41 00:01:39,000 --> 00:01:42,000 So just I guess to get out the the 42 00:01:42,000 --> 00:01:44,000 obvious bit which is what am I actually 43 00:01:44,000 --> 00:01:47,000 doing? What what have I done? So I've um 44 00:01:47,000 --> 00:01:50,000 I joined Mr. as well as the VP of 45 00:01:50,000 --> 00:01:52,000 computational engineering where I'll be 46 00:01:52,000 --> 00:01:54,000 co-leading the AI for engineering side 47 00:01:54,000 --> 00:01:57,000 of things with my very good friend 48 00:01:57,000 --> 00:02:00,000 Johannes who I'll talk a bit uh about 49 00:02:00,000 --> 00:02:04,000 later on and uh and yeah it's um to 50 00:02:04,000 --> 00:02:07,000 summarize it very briefly it's about 51 00:02:07,000 --> 00:02:10,000 developing uh physics AI so foundation 52 00:02:10,000 --> 00:02:12,000 models all the stuff that you know we 53 00:02:12,000 --> 00:02:14,000 talked about in the fluid intelligence 54 00:02:14,000 --> 00:02:20,000 paper um and also the LLM uh side of 55 00:02:20,000 --> 00:02:21,000 things and and to be completely 56 00:02:21,000 --> 00:02:24,000 transparent, one of the reasons I was so 57 00:02:24,000 --> 00:02:26,000 excited to to go to a company like 58 00:02:26,000 --> 00:02:30,000 Mistrial was to be 59 00:02:30,000 --> 00:02:32,000 right there where these leading LLMs are 60 00:02:32,000 --> 00:02:35,000 being developed. And for me, it's an 61 00:02:35,000 --> 00:02:38,000 amazing opportunity to to learn and 62 00:02:38,000 --> 00:02:41,000 contribute and to to make an impact in 63 00:02:41,000 --> 00:02:44,000 that. And that is yeah, super exciting. 64 00:02:44,000 --> 00:02:47,000 I'm not going to talk any more uh on on 65 00:02:47,000 --> 00:02:49,000 this for now. Um also, like I said, 66 00:02:49,000 --> 00:02:52,000 really want this to be more of a sort of 67 00:02:52,000 --> 00:02:54,000 general thing. I'm not here to try and 68 00:02:54,000 --> 00:02:56,000 sell uh the company. You know, you can 69 00:02:56,000 --> 00:02:58,000 there'll be opportunities for that on 70 00:02:58,000 --> 00:03:00,000 more the official sort of uh I guess 71 00:03:00,000 --> 00:03:03,000 mistral channels, etc. Um but this I'd 72 00:03:03,000 --> 00:03:06,000 like to maybe take a step back a little 73 00:03:06,000 --> 00:03:09,000 bit and um and give you idea of sort of 74 00:03:09,000 --> 00:03:12,000 how I got to that and what drove some of 75 00:03:12,000 --> 00:03:14,000 the decisions. Uh but maybe for some of 76 00:03:14,000 --> 00:03:16,000 you who are um like I said earlier on in 77 00:03:16,000 --> 00:03:17,000 your career and this might be some 78 00:03:17,000 --> 00:03:21,000 useful things. So I I guess this started 79 00:03:21,000 --> 00:03:25,000 I did a 4-year mast's engineering um at 80 00:03:25,000 --> 00:03:28,000 Manchester University and probably like 81 00:03:28,000 --> 00:03:30,000 many of you I really wasn't sure what I 82 00:03:30,000 --> 00:03:33,000 wanted to do. I'd got to the end of a 83 00:03:33,000 --> 00:03:35,000 degree 84 00:03:35,000 --> 00:03:36,000 and didn't know what I wanted to 85 00:03:36,000 --> 00:03:38,000 specialize in. Didn't really know what 86 00:03:38,000 --> 00:03:41,000 company I wanted to be at. This was the 87 00:03:41,000 --> 00:03:43,000 days before LinkedIn. you know I am 88 00:03:43,000 --> 00:03:46,000 getting a bit old now just turned over 89 00:03:46,000 --> 00:03:49,000 40 in fact been 41 soon so I guess this 90 00:03:49,000 --> 00:03:51,000 is a while ago and so things are 91 00:03:51,000 --> 00:03:53,000 probably different now uh but back then 92 00:03:53,000 --> 00:03:55,000 uh certainly it wasn't I mean obvious 93 00:03:55,000 --> 00:03:57,000 you know what companies to go to you 94 00:03:57,000 --> 00:04:00,000 know just graduate schemes or going to 95 00:04:00,000 --> 00:04:03,000 um emailing or people etc. It was quite 96 00:04:03,000 --> 00:04:06,000 different uh back then compared to now 97 00:04:06,000 --> 00:04:09,000 and um I had been lucky to do a 98 00:04:09,000 --> 00:04:12,000 placement with a uh racing team. So 99 00:04:12,000 --> 00:04:15,000 Formula 1 was for sure on my mind. I was 100 00:04:15,000 --> 00:04:18,000 lucky enough in um to do my fourth year 101 00:04:18,000 --> 00:04:20,000 project with this team called Iceport 102 00:04:20,000 --> 00:04:22,000 International that was a Formula 2 team. 103 00:04:22,000 --> 00:04:24,000 So the sort of feeder series to Formula 104 00:04:24,000 --> 00:04:27,000 1 at that time called GP2. 105 00:04:27,000 --> 00:04:30,000 and I was doing a project to help 106 00:04:30,000 --> 00:04:31,000 optimize their rear wing in a wind 107 00:04:31,000 --> 00:04:34,000 tunnel and to do some CFD was actually 108 00:04:34,000 --> 00:04:36,000 my first introduction to CFD using I 109 00:04:36,000 --> 00:04:39,000 think Gambit for anybody who is a CFD 110 00:04:39,000 --> 00:04:41,000 person will remember that tool from many 111 00:04:41,000 --> 00:04:45,000 years ago and fluent and I more than 112 00:04:45,000 --> 00:04:48,000 anything the amazing opportunity was to 113 00:04:48,000 --> 00:04:50,000 go to the racetrack. So I actually 114 00:04:50,000 --> 00:04:52,000 shadowed a fantastic engineer called 115 00:04:52,000 --> 00:04:56,000 Gavin. Um, and if if Gavin's listening, 116 00:04:56,000 --> 00:04:58,000 I really should bring him on this 117 00:04:58,000 --> 00:04:59,000 podcast. I'd love to talk to her about 118 00:04:59,000 --> 00:05:02,000 all those years. Uh, and shadowed and 119 00:05:02,000 --> 00:05:04,000 was like a shadow race engineer. I mean, 120 00:05:04,000 --> 00:05:06,000 shadowing in the most thing of it. I 121 00:05:06,000 --> 00:05:09,000 wasn't doing that much. Uh, but I was 122 00:05:09,000 --> 00:05:10,000 there watching how do they set up the 123 00:05:10,000 --> 00:05:13,000 car? How do they decide what to do with 124 00:05:13,000 --> 00:05:16,000 a strategy? I was sat on the pit wall uh 125 00:05:16,000 --> 00:05:19,000 trying to lift up the signs. Uh, it was 126 00:05:19,000 --> 00:05:21,000 an unbelievable experience for someone 127 00:05:21,000 --> 00:05:24,000 21. I think uh at that time also this 128 00:05:24,000 --> 00:05:27,000 for people's interest uh Christian her 129 00:05:27,000 --> 00:05:30,000 was at that time in Formula 2 he owned 130 00:05:30,000 --> 00:05:34,000 the art Grand Prix team um if I remember 131 00:05:34,000 --> 00:05:35,000 correctly I remember seeing him at some 132 00:05:35,000 --> 00:05:39,000 of the races and Teemo Glock was racing 133 00:05:39,000 --> 00:05:41,000 for iceport who later of course had that 134 00:05:41,000 --> 00:05:44,000 controversial thing in Formula 1 that 135 00:05:44,000 --> 00:05:46,000 racing moment you had Bruno Senna of 136 00:05:46,000 --> 00:05:51,000 course Eton Senna's nephew and um Karen 137 00:05:51,000 --> 00:05:53,000 Chandok who now does a lot of commentary 138 00:05:53,000 --> 00:05:56,000 on TV. Uh Luca Degrassi was also around 139 00:05:56,000 --> 00:06:01,000 at that time I think with art and um 140 00:06:01,000 --> 00:06:03,000 and who else was there and yeah several 141 00:06:03,000 --> 00:06:06,000 other uh people that I'm um forgetting. 142 00:06:06,000 --> 00:06:07,000 It was a great experience. It definitely 143 00:06:07,000 --> 00:06:08,000 made me think right I want to get into 144 00:06:08,000 --> 00:06:10,000 Formula 1. Um I think I've told this 145 00:06:10,000 --> 00:06:12,000 story before but I emailed I think 146 00:06:12,000 --> 00:06:15,000 basically every Formula 1 team. Didn't 147 00:06:15,000 --> 00:06:17,000 get any response. I didn't know anybody. 148 00:06:17,000 --> 00:06:19,000 Had no clue. I was probably not even 149 00:06:19,000 --> 00:06:20,000 emailing the right person. I don't think 150 00:06:20,000 --> 00:06:21,000 they even had the email addresses on the 151 00:06:21,000 --> 00:06:23,000 website. It just felt like an 152 00:06:23,000 --> 00:06:26,000 impenetrable thing. [gasps] 153 00:06:26,000 --> 00:06:29,000 And um so for that and linked to various 154 00:06:29,000 --> 00:06:31,000 other things, I'd got into the research 155 00:06:31,000 --> 00:06:34,000 side. I was given the opportunity to do 156 00:06:34,000 --> 00:06:37,000 a PhD and uh as I explained to people 157 00:06:37,000 --> 00:06:39,000 doing a PhD basically was do you want to 158 00:06:39,000 --> 00:06:44,000 carry on being a student but be paid and 159 00:06:44,000 --> 00:06:46,000 do it around potentially formula 1 theme 160 00:06:46,000 --> 00:06:49,000 to thing. So, I took that opportunity 161 00:06:49,000 --> 00:06:52,000 and I'll I'll skip through I won't go 162 00:06:52,000 --> 00:06:53,000 chronologically through every year or 163 00:06:53,000 --> 00:06:56,000 it's going to take forever. Uh but um 164 00:06:56,000 --> 00:07:00,000 during the course of PhD, I finally felt 165 00:07:00,000 --> 00:07:03,000 like I knew what I wanted to go into. I 166 00:07:03,000 --> 00:07:05,000 knew that I could be a CFD or 167 00:07:05,000 --> 00:07:07,000 computational engineering specialist. 168 00:07:07,000 --> 00:07:10,000 That felt [snorts] clearer, which is why 169 00:07:10,000 --> 00:07:12,000 I always I'm an advocate of doing a PhD 170 00:07:12,000 --> 00:07:14,000 because I feel it gives you time to know 171 00:07:14,000 --> 00:07:16,000 what you'd want to do. you could do lots 172 00:07:16,000 --> 00:07:18,000 of other things. Um, in fact, you could 173 00:07:18,000 --> 00:07:20,000 argue the reason I'm doing this podcast 174 00:07:20,000 --> 00:07:22,000 now is because when I was at university 175 00:07:22,000 --> 00:07:24,000 during my PhD, 176 00:07:24,000 --> 00:07:28,000 I ended up being um the like subeditor 177 00:07:28,000 --> 00:07:30,000 of the newspaper for the university. I 178 00:07:30,000 --> 00:07:33,000 got in to journalism. I got into writing 179 00:07:33,000 --> 00:07:36,000 and explaining uh things. Something that 180 00:07:36,000 --> 00:07:38,000 I didn't do in my undergrad and probably 181 00:07:38,000 --> 00:07:39,000 if I'd have left at the undergrad and 182 00:07:39,000 --> 00:07:40,000 gone straight into a job, I probably 183 00:07:40,000 --> 00:07:42,000 wouldn't have done it. The PhD gave me 184 00:07:42,000 --> 00:07:44,000 that sort of time. also gave me a lot of 185 00:07:44,000 --> 00:07:46,000 time to meet people and go to 186 00:07:46,000 --> 00:07:48,000 conferences. I was involved in a lot of 187 00:07:48,000 --> 00:07:51,000 these European workshops, you know, with 188 00:07:51,000 --> 00:07:55,000 EDF Energy, with France, um, with 189 00:07:55,000 --> 00:07:59,000 Airbus, with DR and it was a great, uh, 190 00:07:59,000 --> 00:08:01,000 experience and Alistair Revel who was my 191 00:08:01,000 --> 00:08:03,000 super at the time I owe a lot who sort 192 00:08:03,000 --> 00:08:07,000 of showed me that that world and and but 193 00:08:07,000 --> 00:08:09,000 as my good friend Alistister West and 194 00:08:09,000 --> 00:08:13,000 probably remembers, we I remember saying 195 00:08:13,000 --> 00:08:15,000 to him, if I ever say that I want to do 196 00:08:15,000 --> 00:08:18,000 a posttock, please like you know stop 197 00:08:18,000 --> 00:08:20,000 me. By the end of the PhD I was just 198 00:08:20,000 --> 00:08:23,000 like not interested in academia. I just 199 00:08:23,000 --> 00:08:25,000 wanted to get out. Uh wanted to get a 200 00:08:25,000 --> 00:08:29,000 job. I had no desire to do academia. 201 00:08:29,000 --> 00:08:34,000 And luckily because of the PhD 202 00:08:34,000 --> 00:08:36,000 I'd now I think become a sort of 203 00:08:36,000 --> 00:08:41,000 specialist I guess and I tried again to 204 00:08:41,000 --> 00:08:42,000 get a job in Formula 1. Emailed all the 205 00:08:42,000 --> 00:08:46,000 teams. Basically nobody replied apart 206 00:08:46,000 --> 00:08:50,000 from um two or three teams. Did a few 207 00:08:50,000 --> 00:08:53,000 interviews. Uh a couple I was going into 208 00:08:53,000 --> 00:08:55,000 like aerodynamics which wasn't really my 209 00:08:55,000 --> 00:08:58,000 thing. Went to one which was Lotus. 210 00:08:58,000 --> 00:09:01,000 Applied for an aerodynamics position. 211 00:09:01,000 --> 00:09:03,000 Basically they realized that I wasn't an 212 00:09:03,000 --> 00:09:06,000 aerodynamicist but the the gentleman 213 00:09:06,000 --> 00:09:08,000 Jared 214 00:09:08,000 --> 00:09:12,000 said oh but maybe we could create you a 215 00:09:12,000 --> 00:09:16,000 position to be a CFD engineer 216 00:09:16,000 --> 00:09:18,000 and uh you know we'll do it for 6 months 217 00:09:18,000 --> 00:09:20,000 or and then if you know if you do well 218 00:09:20,000 --> 00:09:24,000 we'll we'll we'll we'll carry on and um 219 00:09:24,000 --> 00:09:28,000 and that was that was my break and so I 220 00:09:28,000 --> 00:09:29,000 have to thank Jared a lot for that. 221 00:09:29,000 --> 00:09:32,000 Jared's now I think the chief engineer 222 00:09:32,000 --> 00:09:36,000 of Mercedes F1 team or I mean he's very 223 00:09:36,000 --> 00:09:37,000 very senior. He basically I think looks 224 00:09:37,000 --> 00:09:40,000 after all of the car development arrow. 225 00:09:40,000 --> 00:09:42,000 Um I haven't seen him for a good few 226 00:09:42,000 --> 00:09:44,000 years. Uh but he he gave me that that 227 00:09:44,000 --> 00:09:48,000 that break and so yeah I left sort of 228 00:09:48,000 --> 00:09:50,000 left university behind and went into 229 00:09:50,000 --> 00:09:52,000 there. Um 230 00:09:52,000 --> 00:09:57,000 then only after leaving academia did I 231 00:09:57,000 --> 00:09:59,000 reflect that I actually missed it and I 232 00:09:59,000 --> 00:10:01,000 realized that 233 00:10:01,000 --> 00:10:02,000 even though I really enjoyed doing the 234 00:10:02,000 --> 00:10:05,000 Formula 1 job 235 00:10:05,000 --> 00:10:08,000 I felt because I've been doing aerospace 236 00:10:08,000 --> 00:10:10,000 and I had so much freedom 237 00:10:10,000 --> 00:10:12,000 and I felt that I was just getting into 238 00:10:12,000 --> 00:10:15,000 a more like dayto-day job there 239 00:10:15,000 --> 00:10:20,000 something didn't fully connect. Uh and 240 00:10:20,000 --> 00:10:21,000 so this is why I sometimes give this 241 00:10:21,000 --> 00:10:24,000 advice now that 242 00:10:24,000 --> 00:10:26,000 don't rule out something. Sometimes you 243 00:10:26,000 --> 00:10:29,000 need time to leave to realize that you 244 00:10:29,000 --> 00:10:32,000 like it. So it was only leaving academia 245 00:10:32,000 --> 00:10:33,000 or leaving after the PhD to go to 246 00:10:33,000 --> 00:10:36,000 industry that made me sort of realize 247 00:10:36,000 --> 00:10:40,000 the benefits of of academia. And so sure 248 00:10:40,000 --> 00:10:43,000 enough, I went back to do a posttock, 249 00:10:43,000 --> 00:10:45,000 the thing that I said I would never do, 250 00:10:45,000 --> 00:10:48,000 leaving Formula 1 behind, but thinking 251 00:10:48,000 --> 00:10:50,000 to myself, 252 00:10:50,000 --> 00:10:52,000 it's not like I'm closing the door 253 00:10:52,000 --> 00:10:55,000 forever, but I feel I want to become 254 00:10:55,000 --> 00:10:58,000 even more of a specialist. I felt good 255 00:10:58,000 --> 00:11:01,000 after my PhD, but ironically, going into 256 00:11:01,000 --> 00:11:03,000 Formula 1 changed my mind. I was like, 257 00:11:03,000 --> 00:11:05,000 ah, this is how industry does it. All 258 00:11:05,000 --> 00:11:07,000 the stuff in academia was great in my 259 00:11:07,000 --> 00:11:09,000 PhD, but I realized I'd been, you know, 260 00:11:09,000 --> 00:11:12,000 not naive, but I hadn't really 261 00:11:12,000 --> 00:11:13,000 understood how the world works. Going 262 00:11:13,000 --> 00:11:17,000 into Formula 1 and industry, maybe got 263 00:11:17,000 --> 00:11:20,000 ah I think I can now come up with a 264 00:11:20,000 --> 00:11:22,000 research direction. I think I know what 265 00:11:22,000 --> 00:11:26,000 my research could fix. So I went back to 266 00:11:26,000 --> 00:11:30,000 do the posttock uh at at Manchester 267 00:11:30,000 --> 00:11:32,000 and uh I actually did it also with CD 268 00:11:32,000 --> 00:11:35,000 Adapt which is now Seammens and that was 269 00:11:35,000 --> 00:11:37,000 another entry point into the software 270 00:11:37,000 --> 00:11:39,000 side. So I actually had a great 271 00:11:39,000 --> 00:11:42,000 opportunity to work with the team at 272 00:11:42,000 --> 00:11:44,000 City of Appco in Hammersmith in London. 273 00:11:44,000 --> 00:11:45,000 Would go down there sometimes and that 274 00:11:45,000 --> 00:11:47,000 again exposed me to the sort of ISV 275 00:11:47,000 --> 00:11:50,000 software world and during the post talk 276 00:11:50,000 --> 00:11:52,000 it was great because then I would really 277 00:11:52,000 --> 00:11:55,000 be going to conferences expanding my 278 00:11:55,000 --> 00:11:58,000 network uh even more and that was when I 279 00:11:58,000 --> 00:12:01,000 had the opportunity to 280 00:12:01,000 --> 00:12:04,000 then eventually move to to Oxford 281 00:12:04,000 --> 00:12:06,000 University. uh that was a great 282 00:12:06,000 --> 00:12:09,000 opportunity where then I really started 283 00:12:09,000 --> 00:12:12,000 to specialize more in HPC 284 00:12:12,000 --> 00:12:14,000 uh and that's sort of where all my high 285 00:12:14,000 --> 00:12:18,000 performance computing sort of uh an IT I 286 00:12:18,000 --> 00:12:20,000 guess you would call it side came in uh 287 00:12:20,000 --> 00:12:23,000 to to to everything and that's the other 288 00:12:23,000 --> 00:12:25,000 thing that I'd advise is that changing 289 00:12:25,000 --> 00:12:28,000 university if you're an academic is good 290 00:12:28,000 --> 00:12:30,000 there's no doubt if you just stay at one 291 00:12:30,000 --> 00:12:32,000 institution your whole time yes you can 292 00:12:32,000 --> 00:12:34,000 do it but I think it does change your 293 00:12:34,000 --> 00:12:36,000 mindset so that's sort of second piece 294 00:12:36,000 --> 00:12:38,000 of advice I guess the first one being 295 00:12:38,000 --> 00:12:39,000 it's okay to switch between industry and 296 00:12:39,000 --> 00:12:42,000 academia and the second one being even 297 00:12:42,000 --> 00:12:44,000 within academia it's probably good to 298 00:12:44,000 --> 00:12:46,000 change institutions to see a different 299 00:12:46,000 --> 00:12:49,000 world to see a different attitude so I 300 00:12:49,000 --> 00:12:52,000 had a great time in the Wes Amore who's 301 00:12:52,000 --> 00:12:54,000 now a professor uh in the engineering 302 00:12:54,000 --> 00:12:56,000 science department was fantastic for me 303 00:12:56,000 --> 00:12:58,000 and um and actually that was my 304 00:12:58,000 --> 00:13:01,000 opportunity then to go to to NASA 305 00:13:01,000 --> 00:13:04,000 because I'd met um Chattin Chatir who 306 00:13:04,000 --> 00:13:06,000 was at that I'm the branch chief at NASA 307 00:13:06,000 --> 00:13:11,000 as he he kind of saw that I had this 308 00:13:11,000 --> 00:13:12,000 experience industrial experience from 309 00:13:12,000 --> 00:13:14,000 Formula 1 which I kind of been bringing 310 00:13:14,000 --> 00:13:17,000 in to my research thinking how can we 311 00:13:17,000 --> 00:13:19,000 solve some of these problems and he 312 00:13:19,000 --> 00:13:21,000 invited me over I think at the same time 313 00:13:21,000 --> 00:13:24,000 Jeff Slopnik uh who has now retired from 314 00:13:24,000 --> 00:13:26,000 Boeing invited me over to to Boeing to 315 00:13:26,000 --> 00:13:28,000 give a talk to them because I think they 316 00:13:28,000 --> 00:13:31,000 had a link with F1 anyway and that 317 00:13:31,000 --> 00:13:34,000 opened up my world to the US And I've 318 00:13:34,000 --> 00:13:36,000 said this before, but this is I've 319 00:13:36,000 --> 00:13:38,000 noticed with academia 320 00:13:38,000 --> 00:13:41,000 there is like quite often 321 00:13:41,000 --> 00:13:43,000 it's almost like a European way of doing 322 00:13:43,000 --> 00:13:44,000 thing in the US. I don't know if this 323 00:13:44,000 --> 00:13:48,000 translates outside of CFD, but everybody 324 00:13:48,000 --> 00:13:52,000 in Europe that there was this 325 00:13:52,000 --> 00:13:53,000 well maybe not Europe, but certainly the 326 00:13:53,000 --> 00:13:55,000 UK. Oh no, I think more Europe. There 327 00:13:55,000 --> 00:13:56,000 was it was coming from like this 328 00:13:56,000 --> 00:13:59,000 incompressible 329 00:13:59,000 --> 00:14:01,000 maybe it wasn't so much an aircraft 330 00:14:01,000 --> 00:14:03,000 design. it was. So I think that's where 331 00:14:03,000 --> 00:14:05,000 open film came out of Imperial very much 332 00:14:05,000 --> 00:14:08,000 with this sort of incompressible 333 00:14:08,000 --> 00:14:10,000 pressure-based segregated type 334 00:14:10,000 --> 00:14:13,000 approaches. Um and that was true at 335 00:14:13,000 --> 00:14:15,000 Manchester with the EDF, a lot of the 336 00:14:15,000 --> 00:14:18,000 codes were like that. And then you'd go 337 00:14:18,000 --> 00:14:19,000 and even if I look at the curriculum 338 00:14:19,000 --> 00:14:22,000 that we had, the examples were with 339 00:14:22,000 --> 00:14:23,000 those sort of formulations. And then you 340 00:14:23,000 --> 00:14:26,000 go to the US and it was all because a 341 00:14:26,000 --> 00:14:28,000 lot of it was done through NASA and 342 00:14:28,000 --> 00:14:30,000 Boeing and aerospace. you were taught at 343 00:14:30,000 --> 00:14:31,000 Stanford and others more in like 344 00:14:31,000 --> 00:14:34,000 compressible density based uh codes and 345 00:14:34,000 --> 00:14:37,000 I remember and like overset meshes and 346 00:14:37,000 --> 00:14:38,000 all these things I'd never come across 347 00:14:38,000 --> 00:14:41,000 before. Uh and that really opened my 348 00:14:41,000 --> 00:14:42,000 eyes and it was the first time I saw 349 00:14:42,000 --> 00:14:46,000 Silicon Valley and ironically I remember 350 00:14:46,000 --> 00:14:48,000 being out in Silicon Valley and thinking 351 00:14:48,000 --> 00:14:49,000 oh wow wouldn't it be amazing to come 352 00:14:49,000 --> 00:14:52,000 and work here. That was 2016 so 10 years 353 00:14:52,000 --> 00:14:55,000 ago. Um 354 00:14:55,000 --> 00:14:57,000 uh and I'm sorry if this is boring 355 00:14:57,000 --> 00:14:59,000 people but I'm going to keep going cuz I 356 00:14:59,000 --> 00:15:02,000 I'm hoping this is also the reason like 357 00:15:02,000 --> 00:15:05,000 if you do a posttock you can also try 358 00:15:05,000 --> 00:15:07,000 and find those placement opportunities 359 00:15:07,000 --> 00:15:08,000 you can find go somewhere expand your 360 00:15:08,000 --> 00:15:10,000 horizon go to a different part of the 361 00:15:10,000 --> 00:15:12,000 world just to see do they do things 362 00:15:12,000 --> 00:15:16,000 differently uh there and then um you 363 00:15:16,000 --> 00:15:18,000 know I I came back to the to the UK kept 364 00:15:18,000 --> 00:15:22,000 you know trying to push things um that 365 00:15:22,000 --> 00:15:24,000 was I started to do also a little bit of 366 00:15:24,000 --> 00:15:28,000 consulting with um with Williams and and 367 00:15:28,000 --> 00:15:33,000 Audi and and uh British cycling team. Um 368 00:15:33,000 --> 00:15:35,000 ended up doing then some stuff for 369 00:15:35,000 --> 00:15:39,000 Formula 1 because um good friend of mine 370 00:15:39,000 --> 00:15:41,000 James James Crook um sort of helped 371 00:15:41,000 --> 00:15:44,000 introduce me. It's a bit of a long 372 00:15:44,000 --> 00:15:47,000 story, but um basically they needed some 373 00:15:47,000 --> 00:15:52,000 help to be doing the um the 2022 uh 374 00:15:52,000 --> 00:15:53,000 regulations at that time and they were 375 00:15:53,000 --> 00:15:56,000 looking to build up like a CFD team. Um 376 00:15:56,000 --> 00:15:58,000 and uh you know, I sort of pitched that 377 00:15:58,000 --> 00:16:01,000 I could do the CFD for them. In reality, 378 00:16:01,000 --> 00:16:03,000 you know, it probably didn't make sense 379 00:16:03,000 --> 00:16:06,000 for them. And so I think they went with 380 00:16:06,000 --> 00:16:07,000 um with Total Sim, you know, who have 381 00:16:07,000 --> 00:16:09,000 become good good friends, but they were 382 00:16:09,000 --> 00:16:11,000 like, well, maybe you could just help us 383 00:16:11,000 --> 00:16:13,000 as like a consultant, as an adviser. You 384 00:16:13,000 --> 00:16:15,000 know, CFD, you can and HBC, you could 385 00:16:15,000 --> 00:16:18,000 maybe stop, give us advice so we know 386 00:16:18,000 --> 00:16:21,000 what we're doing. Uh, and so that that 387 00:16:21,000 --> 00:16:23,000 was an amazing opportunity to go back 388 00:16:23,000 --> 00:16:25,000 into F1 389 00:16:25,000 --> 00:16:27,000 um quite a few years later except now 390 00:16:27,000 --> 00:16:30,000 I'm in London with Pat Simmons, you 391 00:16:30,000 --> 00:16:32,000 know, an icon of the sport, somebody 392 00:16:32,000 --> 00:16:34,000 who, you know, I speak with now 393 00:16:34,000 --> 00:16:35,000 reasonably regularly and someone I 394 00:16:35,000 --> 00:16:37,000 greatly admire as one of the sort of 395 00:16:37,000 --> 00:16:41,000 amazing brains of the the sport who, 396 00:16:41,000 --> 00:16:43,000 yeah, totally gets it, has a very 397 00:16:43,000 --> 00:16:46,000 forward uh looking and it was a bit of 398 00:16:46,000 --> 00:16:49,000 um a shell shock to be uh in F1 in 399 00:16:49,000 --> 00:16:54,000 London with him, Jason uh Somerville 400 00:16:54,000 --> 00:16:56,000 who's now deputy technical director at 401 00:16:56,000 --> 00:16:59,000 Alpine, Simon Dodman who's now one of 402 00:16:59,000 --> 00:17:02,000 the chief engineers at at Cadillac um 403 00:17:02,000 --> 00:17:05,000 and and a couple of other people uh and 404 00:17:05,000 --> 00:17:08,000 uh and sat in London and Ross Brawn's 405 00:17:08,000 --> 00:17:09,000 walking around cuz his office is 406 00:17:09,000 --> 00:17:11,000 basically like right next to you. It was 407 00:17:11,000 --> 00:17:14,000 very surreal to be to be doing it. Um, 408 00:17:14,000 --> 00:17:17,000 and um, it I'm sort of saying this 409 00:17:17,000 --> 00:17:20,000 because I never knew that I would work 410 00:17:20,000 --> 00:17:23,000 for AWS or Nvidia or any of these. I 411 00:17:23,000 --> 00:17:24,000 thought I would always be just like a 412 00:17:24,000 --> 00:17:27,000 CFD person. And and funny enough, during 413 00:17:27,000 --> 00:17:31,000 Formula 1, they sponsored AWS 414 00:17:31,000 --> 00:17:34,000 or AWS sponsored Formula 1. And at that 415 00:17:34,000 --> 00:17:36,000 time, they were like, "Okay, well, we 416 00:17:36,000 --> 00:17:38,000 kind of need somebody who can help us 417 00:17:38,000 --> 00:17:40,000 maybe on this cloud stuff." And because 418 00:17:40,000 --> 00:17:44,000 I got that HPC background from moving to 419 00:17:44,000 --> 00:17:47,000 Oxford, you know, I ended up helping 420 00:17:47,000 --> 00:17:50,000 basically on the HBC side. And uh and 421 00:17:50,000 --> 00:17:52,000 and I've sort of told this story a few 422 00:17:52,000 --> 00:17:56,000 times, but I ended up then also 423 00:17:56,000 --> 00:18:00,000 AWS said, would you mind helping us? Uh 424 00:18:00,000 --> 00:18:02,000 and and you know, Formula 1 were 425 00:18:02,000 --> 00:18:04,000 completely fine with it. So I I was sort 426 00:18:04,000 --> 00:18:08,000 of helping both people at the same time. 427 00:18:08,000 --> 00:18:10,000 um which which was really kind of a 428 00:18:10,000 --> 00:18:11,000 funny situation but it was all done with 429 00:18:11,000 --> 00:18:13,000 complete transparency. It was basically 430 00:18:13,000 --> 00:18:18,000 like hey can you help the AWS people who 431 00:18:18,000 --> 00:18:21,000 weren't CFD specialist help formula one 432 00:18:21,000 --> 00:18:24,000 to get all the CFD going and work with 433 00:18:24,000 --> 00:18:27,000 you know total sim and others and um and 434 00:18:27,000 --> 00:18:28,000 I remember there was like one example 435 00:18:28,000 --> 00:18:31,000 where the code wouldn't scale that we 436 00:18:31,000 --> 00:18:32,000 thought it was because of the 437 00:18:32,000 --> 00:18:34,000 interconnect at the time being slow on 438 00:18:34,000 --> 00:18:36,000 on the cloud you know the whole like EFA 439 00:18:36,000 --> 00:18:38,000 thing getting into too much details for 440 00:18:38,000 --> 00:18:40,000 most people but if you're a HPC 441 00:18:40,000 --> 00:18:41,000 personally you know this is a big 442 00:18:41,000 --> 00:18:44,000 controversy Ethernet Infinibad and they 443 00:18:44,000 --> 00:18:45,000 thought ah it's because it's Ethernet. 444 00:18:45,000 --> 00:18:48,000 So we we basically why should we do this 445 00:18:48,000 --> 00:18:50,000 deal with AWS? You know, the performance 446 00:18:50,000 --> 00:18:53,000 is crap. And uh and turns out I found 447 00:18:53,000 --> 00:18:55,000 out it was uh due to a mesh 448 00:18:55,000 --> 00:18:58,000 decomposition. They were using 449 00:18:58,000 --> 00:19:00,000 uh hierarchical instead of scotch and I 450 00:19:00,000 --> 00:19:02,000 said let's switch to scotch because I'd 451 00:19:02,000 --> 00:19:03,000 done some research in the past that 452 00:19:03,000 --> 00:19:05,000 shows it scales better. Very very 453 00:19:05,000 --> 00:19:06,000 specific thing but basically it showed 454 00:19:06,000 --> 00:19:09,000 that then you could scale a and that 455 00:19:09,000 --> 00:19:11,000 helped a lot the AWS people. I work with 456 00:19:11,000 --> 00:19:15,000 a great guy Lewis Foti on that and and 457 00:19:15,000 --> 00:19:16,000 basically long story short the reason 458 00:19:16,000 --> 00:19:18,000 I'm blabbing on about this is because 459 00:19:18,000 --> 00:19:21,000 that is when AWS said to me would you 460 00:19:21,000 --> 00:19:24,000 like a role at AWS you could be our CTF 461 00:19:24,000 --> 00:19:27,000 specialist because we've seen how you've 462 00:19:27,000 --> 00:19:29,000 been able to sort of translate and help 463 00:19:29,000 --> 00:19:32,000 formula 1 to AWS well we want to go to 464 00:19:32,000 --> 00:19:33,000 all these other engineering companies 465 00:19:33,000 --> 00:19:36,000 could you be our like you know domain 466 00:19:36,000 --> 00:19:38,000 specialist 467 00:19:38,000 --> 00:19:40,000 and there there it was there was my 468 00:19:40,000 --> 00:19:42,000 opportunity to completely change to go 469 00:19:42,000 --> 00:19:44,000 essentially back into industry. It was a 470 00:19:44,000 --> 00:19:46,000 very difficult decision to leave Oxford 471 00:19:46,000 --> 00:19:48,000 at that time. I was still just starting 472 00:19:48,000 --> 00:19:50,000 to get to the point of being offered 473 00:19:50,000 --> 00:19:53,000 like a faculty position. Um it I'm not 474 00:19:53,000 --> 00:19:54,000 going to be I'm not going to lie to you. 475 00:19:54,000 --> 00:19:56,000 It was a struggle. Um I didn't really 476 00:19:56,000 --> 00:19:59,000 succeed in academia. Like I see some 477 00:19:59,000 --> 00:20:00,000 people getting a professorship straight 478 00:20:00,000 --> 00:20:03,000 away. I I don't know. Maybe my research 479 00:20:03,000 --> 00:20:05,000 area just wasn't aligned. I was too 480 00:20:05,000 --> 00:20:08,000 industrial. But um I took that very 481 00:20:08,000 --> 00:20:11,000 difficult choice to to leave and you 482 00:20:11,000 --> 00:20:12,000 know to go to the dark side to go to 483 00:20:12,000 --> 00:20:16,000 industry and there it was I joined AWS 484 00:20:16,000 --> 00:20:19,000 uh an unbelievable experience in the 485 00:20:19,000 --> 00:20:22,000 cloud and it was great because I've 486 00:20:22,000 --> 00:20:23,000 always been a sort of IT person. I did 487 00:20:23,000 --> 00:20:26,000 an IT A level when I was 14. I was a bit 488 00:20:26,000 --> 00:20:30,000 of a geek and so I I got to to do that 489 00:20:30,000 --> 00:20:32,000 and those years at AWS 490 00:20:32,000 --> 00:20:36,000 were probably the most transformative in 491 00:20:36,000 --> 00:20:39,000 terms of working at a you know 100,000 492 00:20:39,000 --> 00:20:42,000 or million person company being in these 493 00:20:42,000 --> 00:20:44,000 commercial discussions understanding the 494 00:20:44,000 --> 00:20:46,000 sales cycle and the technical cycle like 495 00:20:46,000 --> 00:20:52,000 just amazing um and um and throughout 496 00:20:52,000 --> 00:20:54,000 that time I just learned so much I I but 497 00:20:54,000 --> 00:20:56,000 I kept with academia I kept publishing I 498 00:20:56,000 --> 00:20:59,000 kept doing workshops. Um this whole auto 499 00:20:59,000 --> 00:21:01,000 CFD thing I started at Oxford. I carried 500 00:21:01,000 --> 00:21:04,000 on carried on publishing papers. 501 00:21:04,000 --> 00:21:07,000 I've always kept that academic mindset. 502 00:21:07,000 --> 00:21:10,000 Uh and it was a couple of years in that 503 00:21:10,000 --> 00:21:13,000 I I wanted to go back even more 504 00:21:13,000 --> 00:21:16,000 technical and so I moved from the sales 505 00:21:16,000 --> 00:21:18,000 side which is basically where solution 506 00:21:18,000 --> 00:21:20,000 architects were into the product 507 00:21:20,000 --> 00:21:22,000 engineering team at AWS. 508 00:21:22,000 --> 00:21:24,000 and Deborah who's my manager and Ian 509 00:21:24,000 --> 00:21:27,000 colleague were were fantastic for that 510 00:21:27,000 --> 00:21:29,000 and I I then got into like how do you 511 00:21:29,000 --> 00:21:30,000 design 512 00:21:30,000 --> 00:21:34,000 a service I saw this how does the 513 00:21:34,000 --> 00:21:35,000 product team work how does the 514 00:21:35,000 --> 00:21:37,000 engineering team work how do you do 515 00:21:37,000 --> 00:21:39,000 monthly business reviews how do you 516 00:21:39,000 --> 00:21:42,000 design the APIs how do you figure out 517 00:21:42,000 --> 00:21:43,000 the needs what's going to be the revenue 518 00:21:43,000 --> 00:21:44,000 what's going to be the structure of the 519 00:21:44,000 --> 00:21:47,000 code how are we going to make this a SAS 520 00:21:47,000 --> 00:21:49,000 solution what are going to be the 521 00:21:49,000 --> 00:21:52,000 compute requirements EC2 oh my god like 522 00:21:52,000 --> 00:21:54,000 sitting in those calls with Dave Brown 523 00:21:54,000 --> 00:21:56,000 who I really liked. I think he's now 524 00:21:56,000 --> 00:21:58,000 just gone to Meta to lead all their 525 00:21:58,000 --> 00:22:01,000 infrastructure. I just unbelievable 526 00:22:01,000 --> 00:22:03,000 experience. I can't tell you going to 527 00:22:03,000 --> 00:22:05,000 all these big tech companies how much 528 00:22:05,000 --> 00:22:07,000 you learn. Just I can't think I would be 529 00:22:07,000 --> 00:22:09,000 where I am now without having made that 530 00:22:09,000 --> 00:22:12,000 switch to to to AWS. And that's when I 531 00:22:12,000 --> 00:22:14,000 got into machine learning. I was for 532 00:22:14,000 --> 00:22:17,000 years kicking off sort of and I probably 533 00:22:17,000 --> 00:22:19,000 shouldn't talk too much about it openly 534 00:22:19,000 --> 00:22:22,000 just to say there was projects I was um 535 00:22:22,000 --> 00:22:25,000 leading where I really got in to machine 536 00:22:25,000 --> 00:22:28,000 learning uh the surrogate modeling in a 537 00:22:28,000 --> 00:22:31,000 huge way was like so much of my time and 538 00:22:31,000 --> 00:22:33,000 that's actually where the drive ML stuff 539 00:22:33,000 --> 00:22:36,000 was created these open data sets because 540 00:22:36,000 --> 00:22:38,000 I knew that we wanted to develop things 541 00:22:38,000 --> 00:22:40,000 and test things and um we needed open 542 00:22:40,000 --> 00:22:43,000 data and it was my academic IC mind I 543 00:22:43,000 --> 00:22:45,000 could have created that data and kept it 544 00:22:45,000 --> 00:22:47,000 internal but I thought no it's the best 545 00:22:47,000 --> 00:22:49,000 thing for the community that the entire 546 00:22:49,000 --> 00:22:51,000 community benefits it will help us as 547 00:22:51,000 --> 00:22:55,000 well if everything so and I think it has 548 00:22:55,000 --> 00:22:57,000 you put the data out it's now become one 549 00:22:57,000 --> 00:23:00,000 of the most widely used data sets and it 550 00:23:00,000 --> 00:23:01,000 has helped all tech companies to to 551 00:23:01,000 --> 00:23:04,000 advance so uh and that was something I 552 00:23:04,000 --> 00:23:07,000 did with a ML and winter ML and and I 553 00:23:07,000 --> 00:23:10,000 really it taught me I truly believe open 554 00:23:10,000 --> 00:23:13,000 source helps everybody 555 00:23:13,000 --> 00:23:15,000 Um 556 00:23:15,000 --> 00:23:17,000 but 557 00:23:17,000 --> 00:23:18,000 I 558 00:23:18,000 --> 00:23:21,000 felt like if I really wanted to succeed 559 00:23:21,000 --> 00:23:24,000 in the sort of AI for engineering, you 560 00:23:24,000 --> 00:23:28,000 know, do I need to to keep moving on? Is 561 00:23:28,000 --> 00:23:30,000 there other opportunities? And this is 562 00:23:30,000 --> 00:23:33,000 where I had the, you know, great 563 00:23:33,000 --> 00:23:35,000 opportunity most recently with Nvidia. I 564 00:23:35,000 --> 00:23:38,000 joined to sort of do a not a completely 565 00:23:38,000 --> 00:23:41,000 dissimilar role to AWS to be that I 566 00:23:41,000 --> 00:23:45,000 guess domain specialist for for CE now. 567 00:23:45,000 --> 00:23:48,000 I I sort of broadened out and eventually 568 00:23:48,000 --> 00:23:52,000 also to EDA and joining Nvidia 569 00:23:52,000 --> 00:23:56,000 again I felt like another step just 570 00:23:56,000 --> 00:23:57,000 amazing 571 00:23:57,000 --> 00:24:00,000 to see all of the stuff around the GPU 572 00:24:00,000 --> 00:24:03,000 development, the CPU development, um all 573 00:24:03,000 --> 00:24:06,000 the the AI development with like Neotron 574 00:24:06,000 --> 00:24:09,000 and Vix Nemo, the way Jensen runs the 575 00:24:09,000 --> 00:24:12,000 company just like incredible 576 00:24:12,000 --> 00:24:14,000 just like another level, you know. Also 577 00:24:14,000 --> 00:24:18,000 I guess I was slightly biased to AD as 578 00:24:18,000 --> 00:24:19,000 always thinking the cloud going to 579 00:24:19,000 --> 00:24:22,000 Nvidia. Then you you saw the other side 580 00:24:22,000 --> 00:24:23,000 you know that there is this obviously 581 00:24:23,000 --> 00:24:26,000 huge onrem opportunity. I saw now this 582 00:24:26,000 --> 00:24:28,000 in between you know like these neo 583 00:24:28,000 --> 00:24:31,000 clouds etc. Um but also loved getting 584 00:24:31,000 --> 00:24:34,000 into the architecture. It wasn't my main 585 00:24:34,000 --> 00:24:37,000 job but seeing how they make the 586 00:24:37,000 --> 00:24:38,000 decisions on okay what should be the 587 00:24:38,000 --> 00:24:40,000 next GPU? How do we decide this? How do 588 00:24:40,000 --> 00:24:42,000 we, you know, all the benchmark that 589 00:24:42,000 --> 00:24:44,000 goes on the, I can't tell you how many 590 00:24:44,000 --> 00:24:46,000 super smart people there are at NVIDIA, 591 00:24:46,000 --> 00:24:50,000 it's [snorts] nuts. Um, then all of the 592 00:24:50,000 --> 00:24:51,000 CUDA libraries, all the stuff that goes 593 00:24:51,000 --> 00:24:54,000 in at the lower level, you know, the 594 00:24:54,000 --> 00:24:56,000 these enabling libraries that you 595 00:24:56,000 --> 00:24:58,000 basically most people don't know about, 596 00:24:58,000 --> 00:25:01,000 but are what powers all of deep learning 597 00:25:01,000 --> 00:25:04,000 and engineering. [snorts] So, I I was 598 00:25:04,000 --> 00:25:06,000 fortunate to still sort of help with 599 00:25:06,000 --> 00:25:07,000 some of the sales side as well, speaking 600 00:25:07,000 --> 00:25:11,000 to customers. um uh and and the AI side 601 00:25:11,000 --> 00:25:13,000 and the agentic and then working with 602 00:25:13,000 --> 00:25:15,000 some of the chip design team at NVIDIA 603 00:25:15,000 --> 00:25:18,000 on the EDA side. But probably one of the 604 00:25:18,000 --> 00:25:20,000 biggest things I realized at Nvidia was 605 00:25:20,000 --> 00:25:22,000 the partnerships. I totally and I think 606 00:25:22,000 --> 00:25:25,000 this is a Jensen thing the value of the 607 00:25:25,000 --> 00:25:28,000 ecosystem and partnerships and I learned 608 00:25:28,000 --> 00:25:31,000 so much from interacting with cadence 609 00:25:31,000 --> 00:25:34,000 and synopsis and seammens and you know 610 00:25:34,000 --> 00:25:36,000 I've realized the value of the the 611 00:25:36,000 --> 00:25:38,000 ecosystem you know that it's not one 612 00:25:38,000 --> 00:25:39,000 company can do everything it's a 613 00:25:39,000 --> 00:25:41,000 multi-layered 614 00:25:41,000 --> 00:25:44,000 uh cake and um and that that was an 615 00:25:44,000 --> 00:25:47,000 experience that I sort of saw at AWS but 616 00:25:47,000 --> 00:25:49,000 it was turbocharged 617 00:25:49,000 --> 00:25:53,000 uh at at Nvidia and and Uh again I if id 618 00:25:53,000 --> 00:25:55,000 have stayed at AWS no way would I have 619 00:25:55,000 --> 00:25:57,000 the knowledge that I have now. Not this 620 00:25:57,000 --> 00:26:00,000 is nothing against AWS amazing company 621 00:26:00,000 --> 00:26:02,000 just like Azure and GCP and all the 622 00:26:02,000 --> 00:26:05,000 cloud providers but I definitely felt 623 00:26:05,000 --> 00:26:08,000 like I learned so much for Nvidia and to 624 00:26:08,000 --> 00:26:10,000 the point that I was absolutely not you 625 00:26:10,000 --> 00:26:15,000 know looking to leave. I was very happy 626 00:26:15,000 --> 00:26:17,000 um you know that that those 627 00:26:17,000 --> 00:26:19,000 inspirational uh figures there you know 628 00:26:19,000 --> 00:26:23,000 I had I had a great manager. I was so 629 00:26:23,000 --> 00:26:26,000 inspired by people like Ian Buck and um 630 00:26:26,000 --> 00:26:29,000 and and Madison and and Carrie and all 631 00:26:29,000 --> 00:26:31,000 these sort of and Reb and all these like 632 00:26:31,000 --> 00:26:36,000 great leaders, you know. It was so 633 00:26:36,000 --> 00:26:38,000 so good, you know, just felt like every 634 00:26:38,000 --> 00:26:42,000 day I was learning. But I definitely 635 00:26:42,000 --> 00:26:46,000 felt like I could take another step. 636 00:26:46,000 --> 00:26:50,000 That's for sure. Um and when the 637 00:26:50,000 --> 00:26:52,000 opportunity when you know sort of 638 00:26:52,000 --> 00:26:54,000 Johannes was like you've got to come. 639 00:26:54,000 --> 00:26:57,000 You've got to come over. You know this 640 00:26:57,000 --> 00:26:59,000 was Emmy just got acquired by Mistral. 641 00:26:59,000 --> 00:27:00,000 It's like this is it. We can do it 642 00:27:00,000 --> 00:27:02,000 together. Let's let's this is the new 643 00:27:02,000 --> 00:27:05,000 wave, right? It's the Frontier Lab 644 00:27:05,000 --> 00:27:09,000 company um with Physics AI and you know 645 00:27:09,000 --> 00:27:13,000 I really deliberated on it um because as 646 00:27:13,000 --> 00:27:15,000 I said you know I I love Nvidia I still 647 00:27:15,000 --> 00:27:18,000 do today I think it's an amazing company 648 00:27:18,000 --> 00:27:20,000 but I thought this is it. I've never 649 00:27:20,000 --> 00:27:22,000 done a startup, you know, as you saw 650 00:27:22,000 --> 00:27:23,000 from me talking about the history. I've 651 00:27:23,000 --> 00:27:25,000 done academia, gone to AWS, gone to 652 00:27:25,000 --> 00:27:28,000 Nvidia, sold two very big companies. I 653 00:27:28,000 --> 00:27:29,000 thought, oh, maybe going to a thousand 654 00:27:29,000 --> 00:27:31,000 person company. Yes, I can take a bit of 655 00:27:31,000 --> 00:27:33,000 more ownership as well. That's, you 656 00:27:33,000 --> 00:27:35,000 know, obviously I'll be transparent. You 657 00:27:35,000 --> 00:27:37,000 know, becoming a VP, it's it's a nice 658 00:27:37,000 --> 00:27:39,000 thing. It's a nice great progression. Uh 659 00:27:39,000 --> 00:27:42,000 being part of a sort of leadership team, 660 00:27:42,000 --> 00:27:44,000 but it was still my inner academic 661 00:27:44,000 --> 00:27:48,000 scientist wanted to learn. And having 662 00:27:48,000 --> 00:27:49,000 the opportunity to get into this space 663 00:27:49,000 --> 00:27:51,000 of LLMs, 664 00:27:51,000 --> 00:27:55,000 I felt this for sure no matter what 665 00:27:55,000 --> 00:27:57,000 happens, I will learn a lot. It will 666 00:27:57,000 --> 00:28:00,000 always help you know my understanding 667 00:28:00,000 --> 00:28:03,000 and uh and so yeah I I made the move. I 668 00:28:03,000 --> 00:28:06,000 made the move out of a desire to 669 00:28:06,000 --> 00:28:11,000 to learn more I guess and a feel like to 670 00:28:11,000 --> 00:28:13,000 build something. I should say that's the 671 00:28:13,000 --> 00:28:16,000 other thing is that for very good 672 00:28:16,000 --> 00:28:19,000 reasons Nvidia and AWS still stay at the 673 00:28:19,000 --> 00:28:21,000 say not the top level of the stack you 674 00:28:21,000 --> 00:28:24,000 know um they're not making an 675 00:28:24,000 --> 00:28:26,000 application per se it's mainly compute 676 00:28:26,000 --> 00:28:30,000 and libraries uh and um whereas you know 677 00:28:30,000 --> 00:28:33,000 obviously when you go to a company uh 678 00:28:33,000 --> 00:28:37,000 that is directly going to customers then 679 00:28:37,000 --> 00:28:39,000 it is a a slightly different uh 680 00:28:39,000 --> 00:28:41,000 experience and I I kind of fancied the 681 00:28:41,000 --> 00:28:44,000 idea of also building or being at a 682 00:28:44,000 --> 00:28:46,000 company early on in the day, you know, 683 00:28:46,000 --> 00:28:48,000 where things aren't fully defined, where 684 00:28:48,000 --> 00:28:50,000 they need direction and there's the 685 00:28:50,000 --> 00:28:52,000 opportunity to sort of build things from 686 00:28:52,000 --> 00:28:56,000 the ground up. Um, and uh and I yeah and 687 00:28:56,000 --> 00:29:01,000 and then of course I met you know Argy 688 00:29:01,000 --> 00:29:04,000 the the co-founders or all the team 689 00:29:04,000 --> 00:29:07,000 there and just amazing group of people. 690 00:29:07,000 --> 00:29:10,000 Um again for transparency I'm sure 691 00:29:10,000 --> 00:29:11,000 there's amazing group of people open a 692 00:29:11,000 --> 00:29:14,000 nanthropic this is not in a dig at you 693 00:29:14,000 --> 00:29:17,000 know any other uh company but uh but for 694 00:29:17,000 --> 00:29:20,000 me the opportunity to work with with 695 00:29:20,000 --> 00:29:23,000 Johannes who is also an academic he is a 696 00:29:23,000 --> 00:29:27,000 professor we share the exact same 697 00:29:27,000 --> 00:29:30,000 mindset of like we're really fastm 698 00:29:30,000 --> 00:29:32,000 moving and and pushy and and like you 699 00:29:32,000 --> 00:29:36,000 know we know what we want to do but 700 00:29:36,000 --> 00:29:39,000 grounded in science and and rigor and 701 00:29:39,000 --> 00:29:42,000 and openness uh and that sort of pro- 702 00:29:42,000 --> 00:29:45,000 academic um pro publishing pro open 703 00:29:45,000 --> 00:29:49,000 thing is so well aligned uh to myself. 704 00:29:49,000 --> 00:29:54,000 So yeah, here I am. I guess uh the 705 00:29:54,000 --> 00:29:57,000 advice maybe for others is that you 706 00:29:57,000 --> 00:29:59,000 don't know where you're going to go. I 707 00:29:59,000 --> 00:30:00,000 didn't have some grand plan that I'm 708 00:30:00,000 --> 00:30:02,000 going to join an AI company, right? I 709 00:30:02,000 --> 00:30:03,000 didn't even have a plan that I was going 710 00:30:03,000 --> 00:30:06,000 to join a tech company. But looking to 711 00:30:06,000 --> 00:30:08,000 continuously, I guess, push yourself and 712 00:30:08,000 --> 00:30:13,000 be comfortable to take new challenges is 713 00:30:13,000 --> 00:30:17,000 is worth it. And for sure, if you go 714 00:30:17,000 --> 00:30:21,000 into a a large company, a sort of tech 715 00:30:21,000 --> 00:30:24,000 company, I think you can learn so much. 716 00:30:24,000 --> 00:30:25,000 It it doesn't mean that you can't 717 00:30:25,000 --> 00:30:27,000 eventually go back to academia or 718 00:30:27,000 --> 00:30:29,000 whatever, but I I'm still an advocate of 719 00:30:29,000 --> 00:30:33,000 this scientific academic mindset 720 00:30:33,000 --> 00:30:36,000 combined with industrial sort of 721 00:30:36,000 --> 00:30:38,000 application. 722 00:30:38,000 --> 00:30:41,000 So, so that's it. I know I've looking 723 00:30:41,000 --> 00:30:43,000 like a clock here went on for half an 724 00:30:43,000 --> 00:30:45,000 hour and so I hope that wasn't uh well 725 00:30:45,000 --> 00:30:46,000 or maybe it helped you fall asleep. If 726 00:30:46,000 --> 00:30:47,000 you listen to this to fall asleep, maybe 727 00:30:47,000 --> 00:30:49,000 this did its job. But I just wanted to 728 00:30:49,000 --> 00:30:52,000 share um a little bit of my story. I 729 00:30:52,000 --> 00:30:55,000 hope again it just gives you a sense if 730 00:30:55,000 --> 00:30:57,000 you're younger that don't worry if you 731 00:30:57,000 --> 00:30:58,000 don't know exactly where you're going to 732 00:30:58,000 --> 00:31:02,000 go as long as you you know have you know 733 00:31:02,000 --> 00:31:04,000 what you want to do it what makes you 734 00:31:04,000 --> 00:31:07,000 happy what what helps you to continue to 735 00:31:07,000 --> 00:31:09,000 learn don't worry if you can't fully 736 00:31:09,000 --> 00:31:11,000 chart out where you're going to go 737 00:31:11,000 --> 00:31:13,000 there's a funny way that you can go from 738 00:31:13,000 --> 00:31:16,000 place to place and and uh I wouldn't 739 00:31:16,000 --> 00:31:19,000 stress sort of too much uh about it you 740 00:31:19,000 --> 00:31:21,000 don't need to plan out your whole career 741 00:31:21,000 --> 00:31:22,000 right from the 742 00:31:22,000 --> 00:31:26,000 Um but be open to change and and be a 743 00:31:26,000 --> 00:31:27,000 continual 744 00:31:27,000 --> 00:31:30,000 learner. And I guess one thing maybe 745 00:31:30,000 --> 00:31:33,000 okay I would say is and I'm biased. It's 746 00:31:33,000 --> 00:31:35,000 nice if you can become a sort of expert 747 00:31:35,000 --> 00:31:39,000 in something. I feel that that is always 748 00:31:39,000 --> 00:31:41,000 going to help. You know there's nothing 749 00:31:41,000 --> 00:31:44,000 wrong with being a generalist but I do 750 00:31:44,000 --> 00:31:46,000 feel that if you can become a specialist 751 00:31:46,000 --> 00:31:50,000 in an area it probably makes it easier 752 00:31:50,000 --> 00:31:52,000 to chart a career path because you you 753 00:31:52,000 --> 00:31:56,000 are known for a particular thing and I 754 00:31:56,000 --> 00:31:58,000 feel that is probably hard to do without 755 00:31:58,000 --> 00:32:01,000 having done a PhD or to be honest even a 756 00:32:01,000 --> 00:32:03,000 posttock. Um obviously this is not 757 00:32:03,000 --> 00:32:07,000 applicable to everybody but that would 758 00:32:07,000 --> 00:32:13,000 at least be my um my personal uh advice. 759 00:32:13,000 --> 00:32:15,000 There are many other ways of course and 760 00:32:15,000 --> 00:32:18,000 it could be a specialist within your own 761 00:32:18,000 --> 00:32:20,000 area right so it might be that you'll in 762 00:32:20,000 --> 00:32:23,000 formula one but you become known as the 763 00:32:23,000 --> 00:32:25,000 person who's like the AI engineering 764 00:32:25,000 --> 00:32:28,000 expert within your formula one 765 00:32:28,000 --> 00:32:30,000 discipline. I don't mean you have to, 766 00:32:30,000 --> 00:32:31,000 you know, be then going to work for some 767 00:32:31,000 --> 00:32:34,000 tech company or it could be in a in a 768 00:32:34,000 --> 00:32:35,000 software company. You could just be 769 00:32:35,000 --> 00:32:38,000 great uh that somebody who knows product 770 00:32:38,000 --> 00:32:40,000 really well and you you become that that 771 00:32:40,000 --> 00:32:42,000 person. All right. Well, um as you can 772 00:32:42,000 --> 00:32:44,000 tell, I did slip a little bit with a few 773 00:32:44,000 --> 00:32:47,000 episodes because uh all this change has 774 00:32:47,000 --> 00:32:49,000 been a bit u been a bit busy. Uh but but 775 00:32:49,000 --> 00:32:52,000 rest assured there is uh the podcast 776 00:32:52,000 --> 00:32:54,000 will stay on track and I'll try and be a 777 00:32:54,000 --> 00:32:57,000 bit better at getting these episodes out 778 00:32:57,000 --> 00:33:00,000 more more regular. So if I can um help 779 00:33:00,000 --> 00:33:03,000 if people have any questions on sort of 780 00:33:03,000 --> 00:33:07,000 career or um things like that or just 781 00:33:07,000 --> 00:33:09,000 general advice, feel free to leave a 782 00:33:09,000 --> 00:33:11,000 comment. Uh probably on YouTube, I 783 00:33:11,000 --> 00:33:13,000 guess. I think you can do it on Spotify, 784 00:33:13,000 --> 00:33:15,000 but probably on YouTube and I'll try my 785 00:33:15,000 --> 00:33:18,000 best to uh to to answer. So, I hope you 786 00:33:18,000 --> 00:33:21,000 enjoyed that and uh wherever you are, 787 00:33:21,000 --> 00:33:28,000 hope you're doing well. See you. [music] 788 00:33:28,000 --> 00:33:31,000 [music]