2025
Patient Document Research Agent (Mastra AI)
Designed and implemented a config-driven RAG pipeline to ingest and analyze patient documents end-to-end. Built the pipeline around clean interfaces (Strategy + Factory patterns with typing.Protocol) to swap chunking, embedding, retrieval, and ranking strategies without rewriting core logic. Focused on production concerns: reliability, debuggability, and safe iteration as requirements evolved.
PythonRAGLLMsArchitecture
View project →