Google DeepMind

Research Engineer · San Francisco Bay Area

GenAI Reasoning Research Team: agentic coding research.

Meta

Research Engineer · San Francisco Bay Area

GenAI: Post Training. Reality Labs: Robotics.

Google DeepMind

Senior Staff Research Engineer · San Francisco Bay Area

Lead of the Gemini/LaMDA Research Tool Use Group. I developed the tool use technique used in Gemini, later standardized and launched across Alphabet on surfaces including Gemini, Google Search, and Google Cloud.

YouTube

Staff Software Engineer · San Francisco Bay Area

Tech lead and manager of YouTube’s Home Engagement Modeling team, a high velocity, high impact group applying advanced ML techniques at an enormous scale.

• Model scale. Led a multiyear effort to scale our models by a factor of 1,000 on a large fleet of 4th generation TPUs, producing unprecedented increases in daily active users and shifting strategic investment toward scaling models across the organization.
• Knowledge distillation. First team at YouTube to deploy large model distillation, now a key strategic initiative for improving training and serving economics.
• Training infrastructure. Led development of the training framework used by all of YouTube for models trained on user interaction logs. TensorFlow and JAX with first class TPU support. Over 30 distinct model types and thousands of models use it today.
• Modeling techniques. Applied attention based user modeling, AutoML, quantile regression, dense cross networks, uncertainty estimation, bias correction, and more.

Vevo

Senior Machine Learning Engineer · San Francisco

Led development of recommender models and systems.

Yola

Lead Software Engineer · San Francisco