Summary
Kevin Huang is a software engineer with seven years of experience building production ML and AI infrastructure, currently focused on voice agent orchestration at Decagon after contributing to agentic orchestration, vector DBs, and embedding pipelines at Roblox. A Harvard BA/MS in Computer Science with a statistics secondary, he blends research and product instincts—having driven few-shot PyTorch models that improved Roblox search ranking by 15% and worked on multimodal radiology-report generation at Harvard Medical School. He has interned at Meta and Gem where he delivered cost-saving privacy systems and launched the first ML feature respectively, and has a track record of turning prototypes into scalable Airflow/C# services. Comfortable across Python, PyTorch, SQL, and production orchestration, he also teaches reinforcement learning and has led data transparency efforts as president of the Harvard College Open Data Project. Based in San Francisco, Kevin brings an unusually broad mix of applied research, backend engineering, and product-oriented ship-priorities for ML systems. His GitHub bio’s self-deprecating Chinese note hints at a playful curiosity behind a rigorous technical foundation.
7 years of coding experience