At Pinterest, I architected the Pinterest Shopping Assistant, integrating agentic AI workflows and RAG-based conversational discovery across millions of products. The work contributed to a 22% increase in shopping conversion.
I also engineered an LLMOps platform supporting 1,000+ production AI models, reducing model deployment time by 70% and infrastructure costs by 30%. I built retrieval services and evaluation frameworks, and led development of the Home Feed Ranking Platform, which increased user engagement by 18% and recommendation relevance by 25%.
Previously, I built semantic search and recommendation platforms at Google, and developed computer vision and model optimization systems at NVIDIA. My work has included production AI infrastructure, distributed machine learning, and mentoring Machine Learning Engineers.
