Rivian · 2020 – present
Hands-free driving, on every Rivian
I build the scene understanding, motion planning, and control behind Rivian's driver-assistance features, and led several of them from zero to launch. These systems decide where the lanes are, which one the vehicle is in, and the path it should take, then steer it there smoothly. They run on every Rivian platform, and Universal Hands-Free now works on more than 3.5 million miles of roads in the US and Canada. I also drive my own R2 every day, and it's always great to test my own software.
Universal Hands-Free
Automated Lane Change
Large driving model
Scene understanding
LKA · LDW · AEB · FCW
Patent pending
- Large driving model: my most recent work is on post-training and guardrail design for Rivian's large driving model.
- Zero-to-one launches across the planning and control stack: the lateral controller, scene-understanding and lane-geometry engines, the move to in-house perception, and HD/SD-map integration.
- Lane-level localization: a Bayesian filter that fuses bird's-eye-view lane topology, SD-map, and barrier cues with turn-signal intent to estimate lane count and the ego lane with calibrated confidence.
- Motion planning and trajectory optimization powering hands-free driving and automated lane changes.
- In-house safety features (lane keeping, lane departure warning, automatic emergency braking, forward collision warning) from prototype to production-intent validation.
- Technical lead: set architecture and roadmap, ran design and code reviews, and mentored engineers on planning, controls, and V&V.