Frank May

I’m a Senior Full-Stack AI Engineer focused on turning advanced LLM and agentic ideas into production systems that teams can trust. At Abridge, I architected an end-to-end AI platform for clinical conversation intelligence, transforming multimodal inputs (audio, transcripts, and EHR context) into structured, auditable outputs at enterprise scale.

I lead orchestration and workflow design—coordinating models, tools, APIs, and contextual memory for multi-step reasoning and dynamic execution. I also built agentic AI frameworks (LangChain, LangGraph, LlamaIndex, AutoGen-style systems) with retrieval augmentation (RAG), persistent session-level memory, and evaluation/feedback loops for reasoning quality, hallucinations, factual grounding, and task success.

I’ve engineered the underlying runtime and data layers—high-performance inference and GPU-backed infrastructure (AWS, Kubernetes), vector databases (pgvector/Pinecone), embeddings, hybrid search, and re-ranking. Earlier, I built end-to-end product and platform features at Instacart and large-scale distributed software at Google, and I consistently emphasize MLOps, observability, reliability, and mentorship through scalable design best practices.