Qasim Ali

Hi! I’m Qasim Ali, a Founding Research Scientist at GSI (Grounded Superintelligence). We’re building human foundation models: models trained on egocentric human data at global scale to bring general-purpose physical intelligence to robots 🤖.

I believe strong representations are the most important ingredient for learning, and that World Models 🌎 might be the best way to build them. I see robotics as the best testbed for measuring intelligence. If we can solve it, we’ll have a learning algorithm efficient enough to scale anywhere. LLMs haven’t cracked robotics, which tells me they aren’t the last learning algorithm 🧠.

Before GSI, I was a graduate student at the University of Waterloo, advised by Alexander Wong and Javad Shafiee, who are now both directors at Apple. My research there covered world models, generative models (diffusion and flow) and reinforcement learning.

I come from an embedded software and robotics background. In my Mechatronics Engineering undergrad, I built a control algorithm for vehicles navigating V2I-connected intersections for the EcoCar Contest, and built a Drone. I also did a 16-month internship at Magna Electronics’ ADAS division, working across embedded software platforms to get self-driving features running on vehicles.

If you’re into AI research, robotics, or just want to chat about the future of physical intelligence, feel free to email me!