Introduction
I found a go kart in the trash with the illusion of building an electric go kart and I ended up leading an autonomous IndyCar team to push the limits of driverless vehicle dynamics at 160mph (250kmh) as the vehicle state estimation lead winning the podium at 2nd and 3rd place.
- B.S. Computer Engineering @ UC San Diego
The 9am-5pm @ Google and 6pm-2am @ IAC
-
Googler during the day
Vehicle State Estimation Lead during the night
- Implemented a hyper-precise sensor fusion stack achieving 10cm accuracy.
- The perception-based pose estimate combines LiDAR, camera, and kinematic pose estimations for a 160mph non-linear dynamic system.
- Traveled around the country to compete at the highest level in Las Vegas, Fort Worth, and Indianapolis against PhD programs from Germany, Italy, South Korea, Canada, Emirates, and the U.S.
--- ---
 |
 |
|---|---|
 |
 |
Building deep learning systems at the Google scale
-
For the launch of Immersive Navigation, I recruited the best talent and led a team of 4 engineers.
- Trained, deployed, and globally scaled a vehicle dynamic learning model in 9 countries to reward beautiful visual lane guidance for +13M miles affecting 212M users.
- Achieving above 90% precision and 85% recall across all road types.
- I was honored with the Geo 2025 Tech Impact Award and a performance review of a Transformative Impact (top 5% of Google).
1st author of a U.S. patent.
-
Drawing a novel connection between medicinal research (schizophrenia & human disease network) and Google Maps. It turns out the build dependency graph of Google Maps is not dissimilar to a biological nervous system.
-
Cracking open a decade old app using unsupervised learning to detect functional modules to improve system health by 10x in detangling the build graph into neural modules.
-
I was also honored with the Geo 2024 Innovator Award and the Geo 2024 Tech Impact Award.
Bringing personal maps data into AskMaps, AI Mode, and Gemini
- Leading a team of 7 engineers to build an agentic world simulation to forecast personalized experiences to 2M DAU in the US & IN.
- Humanity first. Building an AI assistant that is maximally helpful and personal at the Google scale is ethically complex. Privacy is not a compromise, its a design decision.