Research Data Scientist - AI System Knowledge and Information Engine (ASKIE)
February 2026–present · New York City, NY
Joined shortly after Gemini 3.1 Pro's launch; tenure spans the introduction of Gemini 4 Argon.
- Lead the Work Done by AI team, designing and owning Google-wide metrics of AI contribution, primarily focused on Antigravity coding agents powered by multiple models, including Gemini 4 Argon.
- Own the development of grading algorithms for Google Antigravity coding-agent trajectories, measuring effort-equivalent AI contribution, correctness, complexity, and impact; completed trajectory analysis intended for use by Google DeepMind and other internal teams.
- Developed a modular statistical-reporting and visualization skill in Gemini's internal codebase, used by 500+ Google employees and integrated into team data-science and organizational insight agents; supports standalone use and composition within larger skill suites, and serves as an internal example of effective skill design.
- Conduct cross-product studies of AI adoption, employee behavior, and productivity across Google, applying double machine learning and continuous-treatment difference-in-differences to inform development priorities for next-generation coding agents. Analyze hundreds of billions of first-party coding-agent and LLM usage records.
Research Data Scientist - Google Hardware (Pixel)
September 2024–January 2026 · Chicago, IL
- Built an LLM sentiment-analysis pipeline using LLM-as-judge, SFT, and RAG; achieved 99.5% accuracy and informed hardware KPI targets and RMA analysis.
- Developed a Pixel diagnostic pipeline using LoRA and RLHF; improved classification accuracy from 40% to over 90% (0.95+ recall) and increased throughput by 300%.