I'm a technical program leader who has shipped 9 frontier models and agentic systems at Microsoft AI Superintelligence on an 8-week release cadence, and led generative AI strategy at Google Shopping. I run release, evaluation, and safety programs across reasoning models and computer-use agents.
At Microsoft AI, I own the product portfolio for the small language model and agentic AI programs — from Phi through Fara1.5 — and built the org's model development, evaluation, and release pipeline: a single readiness path every model now ships through. That means driving CUA, RL, Responsible AI, Legal, GTM, Sales, Foundry, and Marketing to each launch.
Before Microsoft, I spent nearly three years at Google as Product Strategy & Research Lead on Shopping's generative AI work, defining strategy for the Search Generative Experience and owning the team's genAI roadmap.
I hold an M.S. in Human-Computer Interaction from Georgia Tech and a B.Tech in Chemical Engineering from NIT Karnataka. I'm driven by the belief that frontier AI should be both capable and responsibly released.
Outside of work, I'm a creative at heart — an avid reader (55 books in 2022!), singer, painter, and baker who experiments with recipes like lavender mint cake.
Worked in consulting before transitioning to tech, gaining experience in strategy and problem-solving across industries.
Scalable, evolving training environments for computer-use agents — deepening the task distribution agents learn from.
Computer Use EnvironmentsThe Fara1.5 portfolio — a three-tier 4B/9B/27B family whose 27B model reached 72% on Online-Mind2Web, surpassing OpenAI Operator and Gemini 2.5 Computer Use.
Agentic AI Computer UseA 7-billion parameter computer-use model that beats GPT-4o at roughly 1/12th the inference cost, with a Critical Points framework requiring human approval before irreversible actions.
Agentic AI Computer UseThe 14B reasoning model that matches DeepSeek-R1 (671B) on AIME 2025, establishing Microsoft's position in efficient reasoning.
Reasoning Language ModelsInvestigation of inference-time compute scaling methods across reasoning tasks, exploring how additional compute at inference improves performance.
Scaling ReasoningA framework for synthetic data creation using agentic flows — its 25M instruction pairs became the training foundation for every subsequent Phi model.
Synthetic Data AgentsBuilt the org's single readiness path for model development, evaluation, and release — replacing per-launch coordination with a repeatable process every model now ships through.
Microsoft · ProgramDrove the Phi reasoning program from concept to launch — Phi-4 (14B) through Phi-4-reasoning-vision (15B), matching DeepSeek-R1 (671B) on AIME 2025.
Microsoft · ReasoningDefined Microsoft's computer-use agent strategy — Fara-7B beating GPT-4o at 1/12th the cost, then the Fara1.5 4B/9B/27B portfolio hitting 72% on Online-Mind2Web.
Microsoft · AgentsDesigned the safety framework for computer-use agents: human approval required before any irreversible action the agent would otherwise take on its own.
Microsoft · SafetySole PM for the agentic product surfacing this work — ran market research, drove development with research and engineering to product-market fit, and owned GTM through launch.
Microsoft · ProductDefined product strategy for Shopping's SGE, the team's top AI initiative — 7% lift in user satisfaction and 3% lift in impressions across Google Search.
Google · GenAILaunched Virtual Try-On as part of owning Shopping's generative AI roadmap, alongside the LLM landscape study that set the team's AI product direction.
Google · GenAIWinner of the Executive Challenge at Microsoft's flagship internal hackathon, competing against teams worldwide.
AwardInvited speaker at Microsoft's Machine Learning and Data Science conference, and proposal reviewer for MLADS+RAI submissions.
Speaking · ServiceI'm always interested in discussing frontier model release and evaluation, agentic AI, or potential collaborations. Feel free to reach out.
The Design of Everyday Things
"The best way to predict the future is to invent it."— Alan Kay