At Meta, I architected production AI/ML systems for Facebook and Instagram recommendation surfaces. I advanced feed and Reels recommendations with two-tower retrieval, learning-to-rank, and multimodal embeddings, while improving feature freshness latency by 60%.
I developed generative AI capabilities around Meta's Llama ecosystem, combining retrieval-augmented generation, vector search, and agentic orchestration. I also optimized production inference, lowering p95 latency by 30%, increasing serving throughput by 2.1x, and reducing compute cost per prediction by 20%.
At Realtor.com, I led personalization for home search and discovery, increasing search-result relevance by 10%. Earlier, I engineered banking data pipelines at JPMorganChase and developed property valuation and mortgage technology applications at CoreLogic.
