Microsoft · GitHub Copilot
Senior Software Engineer, Applied Science and Model Evaluation
January 2026–Present
- Built cross-platform grading and run orchestration for an LLM evaluation platform supporting 300K+ submitted runs and 10M+ task evaluations each month.
- Implemented privacy, security, and benchmark-integrity graders producing 10M+ pass/fail judgments monthly, plus a self-service grader registry and reproducible release pipelines.
- Developed a Harbor-native benchmark with 58 real-world Windows tasks and productionized analysis generating 2,600+ full-run reports per month.