Led a team of 75 Data Scientists for improving Gemini Advanced in collaboration with Google Bard's ICE team. Owned model evaluation and error discovery processes. Generated metrics on model correctness and compared model versions to achieve steady improvement. Investigated statistically significant error patterns and proposed fix proposals with research teams. Recommended training data policies for SFT, led high-quality throughput RLHF data generation, and integrated RLMF processes with synthetically generated data.

Senior ML Scientist at Omnisight

August 1, 2019 - February 1, 2021

Developed deep neural networks for outdoor advertisement insights (human detection, face recognition, and pose estimation). Optimized deployment on NVIDIA Jetson with TensorRT, and deployed a real-time multi-object tracking system to deliver interactive analytics.

ML Scientist at Science Wave Capital

September 1, 2018 - March 1, 2019

Created predictive models for forecasting stock returns and 5-day alpha values for a market-neutral portfolio using LightGBM, XGBoost, and neural networks; performed feature engineering to adapt signals to volatile market conditions; supported daily reallocation of a $150M leveraged portfolio, integrating model predictions with Markowitz optimizers for optimal results.

AI Engineer at Infotech

November 1, 2016 - September 1, 2018

Led AI efforts in the TEMSA MD9 electric autonomous bus project, focusing on sensor data processing and fusion from LIDAR/Stereo Cameras for object detection and tracking. Implemented ML models for government R&D, including indoor localization using smartphones and driver risk classification using telematics.