Ron Thalanki

ML engineer · Meta Reality Labs

I build computer-vision detectors, and the data engines behind them, for Codec Avatars, Meta's photorealistic, drivable avatars for AR/VR.

My focus is data-centric ML: getting more out of large vision models through better training data (synthetic generation, auto-labeling, and curation) rather than just bigger models.

Computer Vision Foundation Models Synthetic Data Auto-Labeling Data-Centric ML

Outside work, I tinker on fun projects like Passive Alpha and sports analytics (World Cup surprise rankings). Off-screen: hiking, baking, and salsa dancing.

Studied EECS & Business at UC Berkeley.