Jan 2020 – Present · 6 yrs 5 mos · Greater Seattle Area · Hybrid

• Architected end-to-end retrieval-augmented generation (RAG) infrastructure, grounding AI responses with live search indexes, answer content, and partner knowledge bases for internal teams and 3P integrations (OpenAI, X.com, Meta).
• Spearheaded Responsible AI compliance in Bing Copilot Search by embedding offensive/adult classifiers and collaborating with Trust & Safety, Legal, and Policy teams to enforce guardrails, prevent misuse, and uphold brand integrity.
• Unified UI/UX across platforms through designer tokens and common controls, aligning with global design systems to modernize Bing’s look and feel and boost click-through and engagement metrics.
• Reduced backend latency by 15 ms on Bing’s critical search path—translating to ~$15 M in annual revenue savings—via deep full-stack optimizations, asynchronous data flows, and fine-tuned caching strategies.
• Built real-time pipelines to suppress offensive or reported AI content, editorially rank answer segments, and stream tokens for Copilot RAG prototypes, lifting query-level engagement and reducing customer dissatisfaction.
• Championed developer productivity enhancements: integrated MCP servers and GitHub Copilot into Azure DevOps pipelines to auto-generate unit tests, syntax corrections, and CI/CD deployments—cutting cycle times by 50% for 500+ engineers across 30+ teams.
• Designed real estate search grounding pipelines aggregating Zillow and Redfin data, developed multi-language PCFG models for advanced query parsing (price, year, amenities), and engineered POI clustering on Bing Maps to elevate property discovery at every zoom level.