At Microsoft, I own Unified Product Search initiatives across Dynamics 365 Commerce. My work includes swatch integration, catalog backfilling, localization, and catalog-specific pricing and data correctness.
I designed recovery for denormalized publishing and implemented deletion orchestration across commerce data services, Retail Server, and Azure Search. Through caching and workload reuse, I reduced median publishing runtime by 17.46% in a local synthetic benchmark.
At Microsoft Research, I developed architecture and client-web features for Kahani, a visual storytelling tool using Python, large language models, and Stable Diffusion XL. I also conducted a user study with 21 participants and wrote a research paper for UIST 2024.
Earlier, I worked on notification and listing features at Nbyula, contributed to pricing and automated testing work at Amazon, and built production-alert analysis features during my Google internship. I also build backend projects including a distributed job scheduler, a payment orchestrator, an event-driven e-commerce system, and a ride-hailing platform.
