Domain-oriented data ownership and contextual memory for adaptive RAG - not just a vector search.
Architected a distributed data mesh retrieval platform with domain-oriented data ownership, semantic vector search, and contextual memory to deliver adaptive RAG across distributed structured and unstructured data. Engineered hybrid retrieval pipelines combining embeddings, graph retrieval, temporal context, and personalized memory for context-aware querying, recommendations, and conversational intelligence. Developed a real-time streaming AI interface with contextual citations, persistent memory, and scalable self-serve data infrastructure using FastAPI, Redis, PostgreSQL, and FAISS deployed via Docker on AWS.
Next.jsTypeScriptFastAPILangChainLlamaIndexPostgreSQLRedisFAISSNeo4jAWSDocker
AWS