Principal Engineer (Director) at Google DeepMind (and previously at Google Brain). Leading a team and conducting research on large multi-modal foundation models with deep visual reasoning and memory capabilities.

Worked at Meta / Facebook AI Research (FAIR) Accel and started new projects in interactive continual learning for building multi-modal AI assistants for AR devices.

For six years, I spearheaded cutting-edge research in machine learning, perception, and prediction for Google's self-driving car project, now known as Waymo. For the last two years, I led the core deep learning research team to identify and undertake ambitious long-term research in end-to-end perception and planning. I have helped lead, build, and deploy three-generations of camera and laser perception systems on the car with high quality real-time deep neural networks. I have extensive experience in writing large-scale infrastructure from scratch for training various machine learning models (neural nets, decision trees, cascades), and have written the first onboard neural net inference engine deployed on the self-driving car in 2013.

At Google, I worked on researching and building state-of-the-art machine learning perception and prediction systems for the self-driving car project, and on large-scale 3D computer vision in Google Maps (Street View, Google Earth) before that.

Oct 2016 - Principal Engineer/Director Level 8

Mar 2013 - Senior Staff Software Engineer Level 7

Apr 2012 - Moved to Project Chauffeur (Google Self-driving cars).

Oct 2009 - Staff Software Engineer Level 6

Oct 2007 - Senior Software Engineer Level 5

Oct 2006 - Joined Google Maps as Software Engineer Level 4

I worked on 3D computer vision (structure from motion), image correspondence (optical flow), and grammars for human action recognition.