Looking for a job

Michael Yang

@michaelyang3

Senior Machine Learning Engineer at Lyra Health, improving care search-to-booking conversion 18% with personalized ranking.

United States

Message

At Lyra Health, I led machine learning for Natural-Language Care Search, using dense retrieval and personalized learning-to-rank to improve search-to-booking conversion by 18%. I also built clinical safety and provider eligibility constraints into rankings serving 150K+ queries daily.

At Amazon Web Services, I worked on Amazon Bedrock, leading development of Knowledge Bases retrieval and ranking components through GA launch. I also designed inference optimizations that reduced customer inference costs by ~35% and improved throughput 2.5×.

At edX, I designed a real-time adaptive recommender that improved recommendation CTR by 24% and course completion by 8%. I used contextual bandits to lift new-course enrollment 15% and built streaming pipelines processing 200M+ learner events per day.

Experience

Work history, roles, and key accomplishments

Current

Senior Machine Learning Engineer

Current

Lyra Health

Aug 2025 - Present (1 year 2 months)

Led ML for Natural-Language Care Search, a large-scale recommendation and ranking system, improving search-to-booking conversion by 18% over keyword search. Designed candidate generation and personalized ranking stages, and established an end-to-end ranking evaluation stack.

Dense Retrieval Faiss Learning To Rank Gradient Boosting Pairwise Objective Constrained Ranking Clinical Safety Rules HIPAA A B Testing Counterfactual Evaluation Propensity Scoring Mentorship

Senior Machine Learning Engineer

Amazon Web Services (AWS)

Aug 2019 - Jul 2025 (5 years 11 months)

Core engineer on Amazon Bedrock, leading development of Knowledge Bases (RAG) retrieval and ranking components from design through GA launch. Designed model inference optimization layer, scaled distributed serving on GPU fleets, and sustained 99.95% availability on a Tier-1 service.

RAG Embedding Based Retrieval Reranking Relevance Scoring Continuous Batching KV Cache Optimizations GPU Quantization INT8 FP8 Dynamic Request Batching Routing Canary Testing Rollback Tooling Mentorship

Data Scientist

edX

Jul 2016 - Aug 2019 (3 years 1 month)

Designed a real-time adaptive recommender personalizing course and content suggestions, improving recommendation CTR by 24% and course completion by 8%. Deployed contextual bandits for real-time exploration/exploitation and built streaming pipelines processing 200M+ learner events/day.