At Microsoft AI, I build LLM-driven systems that surface credible health information across Copilot and Bing.
I operationalised the National Academy of Medicine credible-source framework as an LLM classifier, co-designing prompts with clinicians and shipping a production pipeline that expanded the credible-source set by roughly 10x.
I also designed a forward-simulator pipeline for synthetic wearable data from clinician-authored patient seeds, alongside an n8n sourcing workflow that orchestrates LLM calls and containerized Python services for credible health content.
Previously at Microsoft, I built backend systems for TakeLessons India and Bing, including a lead feature that increased converted leads by about 60% and infrastructure supporting Bing experiences at roughly 1.4M peak daily-active users.
