Ali Agha, a leader in AI and autonomy algorithms, and his experience in raising $400M. Watch the full video — free, no account needed.
Ali Agha, a leader in AI and autonomy algorithms, and his experience in raising $400M for FieldAI. The discussion highlights his journey from NASA to founding a deep tech startup focused on AI-driven brains for autonomous robots.
obsession with the with what actually would enable the solution to be deployed at millions and bake that into your framework both on the business side and the technology side. I think it would be the most critical thing to to build the type of a startup that we are building, a deep tech startup. >> Alrighty, hello everyone and welcome to the Deal Maker Show. So today we have an awesome founder. We're going to be talking quite a bit about robotics. So I hope that you're ready. You're ready for his journey. Incredible what they he's building now with his company. You know, they've raised a quite a bit, you know, 500 million plus and we're going to be discussing the journey, you know, also how he went from maybe more the academics to really implementing and and and having that technical depth, you know, how that looks like when you apply it really to the execution. And quite a rocket
ship that they're building. So again, brace yourself for this really inspiring conversation. So without further ado, let's welcome our guest today, Ali Aga. Welcome to the show. >> Awesome. Great to be here, Alejandro. >> So originally, you were, you know, uh in this snowy region in the world and and essentially you had this fascination for math and and and from the very early days. So that fascination, where did it come from? >> Yeah, I mean, as far as I remember, um yeah, I always loved math. I always loved I mean, tinkering with tools and engineering and put a few batteries and and make a motor sort of start spinning um and you know, always loved playing with the toys that can do something a little bit on their own, right? Even if it's simple, um but it was always with me and I think over time it just grew as I learned more. Oh, I can actually code and bring
something to life and have these machines actually do something, have piece of software do something and create something. It just It just grew more over time more and more in me and um all of that I think is is led to sort of where I am now. I mean I I all of that passion led me join um and in my undergrad robotics team here in front of the early years of my undergrad building robots and you know, programming them to from following a line at the very first ever robot I made like two decades ago all the way to small little soccer robots that are kicking a orange ball in RoboCup like competitions and and just make them you know, smarter and smarter but but that's that's how it all all started for me. >> So, talk to us too about coming to the US because coming to the US you know, via Texas where you were doing your PhD, you know, was a was quite a pivotal moment. >> Yeah, I
mean um like I mentioned I I in my undergrad and masters I was doing quite a bit of robotics. So, I I did my undergrad and masters in in sort of a program between electrical engineering and computer science and um as I mentioned working on these RoboCup competitions started to uh instill in me how much work needs to be done on the theoretical side, how much work it needs to be done on the on improving the algorithms for these platforms to actually um show performance that handle the the real world. So, I in my masters even before I come to US now I moved to a bit more sophisticated platforms. It's still within robot robotic competitions RoboCup competitions, but in a different league where we were in building rescue robots. These tracked robots with flippers, very complex platforms going over step field unstructured environments and was a lot of fun and I started to appreciate how
critical it is to have robust methodologies to understand the world, understand the state of the robot, where it is in this world, how it make decision, how to do those in a safe way, how it learns over time and um we actually from that sort of small school on the other side of the world we came and won their best mobility in in RoboCup competitions in in Atlanta, Georgia. This is I I believe it was 2006-2007 time frame. And and and from there as you alluded to I I'm that appreciation to the fact that we need better technologies, better theoretical frameworks for these robots to show a robust behavior. I moved to to Texas to do my PhD in in in computer science and uh in my PhD I actually took I think more courses from math department than than any other other department and that really was again to go really deep in elements of the solution that I believed robots required to to operate
reliably that were beyond the the the theories and beyond the methodologies that were in the textbooks, right? When you go to the the real world You are dealing with linear system or Gaussian noises or uh you know, a a lot of assumptions that are incredibly important to learn and build your your your your basics, but they don't directly apply. You need to take them, you need to learn them, but then go beyond those to reach a point that develop a solution that actually survives all the off-nominal cases, all the complexities of the real world. And I think that appreciation to the deployments in these competitions, that appreciation to uh deploying the robots in the real world helped me to appreciate in an ironic way a depth of the theory. So, I in my PhD I went quite deep in in in theory, went quite deep in uh probability theory, stochastic differential equations, understanding the risk,
quantifying the the the uncertainty and what can go wrong and what are the delta and difference between computational models, your assumptions, and what you actually face in the real world. And and um we did some very interesting work. I wrote some seminal sort of papers, introduced a a few new uh frameworks there, which made the robots smarter, made them uh accomplish tasks that um were were beyond what was possible in that sector that I was working on uh uh prior prior to those work. And uh it was was pretty fun time. >> Which it was quite a shift to like um you know, going more from the academia because then you did MIT to all of a sudden you find yourself in in places like Qualcomm or or NASA. You know, at what point did you you know, think that perhaps having that shift um and change course of action, you know, in your career was the way to go? >> Yeah, I I don't see it
really as a as a really a shift. I think it was a natural progression. So, post my PhD, I moved to to MIT. I was a I was a postdoc there. Uh I was working with an incredible group of people, and we I was now starting to take what I had done at my PhD and from a a single robot autonomy to a coordinated operations across a team of robots. Flying platforms with the ground platforms coordinating and together in a decentralized manner taking action. So, that was a deeper level to now build multi-agent autonomy, multi-agent AI. And um one of the incredible things at that period was I basically met my co-founder. He was one of the best students that as a postdoc I was supervising, and he um he he went for for DeepMind building, uh you know, models and and and pushing uh on the uh foundation model and machine learning dimensions. And for me, uh I moved from MIT to Qualcomm because I was getting
fascinated by making very small models, the autonomy that runs on resource constraint platforms. Qualcomm at the uh moment were were getting into robotics and drones and self-driving car and was uh were were looking into putting algorithms on their their their their processors. Uh so, I I moved there, um contributed to bringing perception and autonomy on their chip. We did the first CES show uh for them on the robotic and and and drone side. It was incredible, and um uh and it it was still, I would say a a very continuous progression of what I was doing, but now on a resource constrained platform. And that caught the attention of of NASA as well. At that at the moment, Mars helicopter was not a publicly announced mission. Uh but um NASA was in conversations with with Qualcomm on exploring usage of their lightweight low-power processor on Mars helicopter. And that's how I transitioned. I
I I I moved to NASA. I was fascinated with with this once-in-a-lifetime opportunity uh and to >> And what what was the level of intensity there? Because I I believe it was quite intense. So so what level of intensity are we talking about? >> Yeah, I think my my work in NASA maybe has has two parts, right? So on on sort of the management track, I was uh a group supervisor in aerial mobility group. Um and and and later in the in the perception group. And as as you might know, you know, Mars helicopter flew on Mars with Qualcomm chip on it. So that was incredibly fun to watch and see this massive incredible team in in NASA JPL, how they operate with the maximum of course level of intensity. You can't move Earth and Mars relative location, right? So that that everything is moving in in the universe. And the launch date is is the launch date. Everything needs to be done and and
completed. So I I…