Jasman User

I'm an AI Engineer with 4+ years of experience designing, building, and deploying production RAG pipelines and multi-agent LLM systems. I currently build a fully on-premise, air-gapped RAG platform for regulated document intelligence — hybrid retrieval (dense + BM25 + RRF), LangGraph multi-agent orchestration (router → tool-executor → responder), and LLM inference on constrained GPUs (llama.cpp, vLLM) with zero external API dependencies for data-sovereignty compliance.

My background spans FastAPI backend engineering (async PostgreSQL, Celery/Redis task queues, JWT auth, Docker), vector-DB migrations (ChromaDB → Milvus/Qdrant), and open-weight model selection for constrained hardware. Earlier in my career, I built AI-powered analytics dashboards and automation pipelines that cut manual processing time from hours to minutes, and led 35+ hardware development projects in TV R&D, building Python automation frameworks that reduced manual monitoring workload by 70%.

I collaborate closely with cross-functional teams to translate business and compliance requirements into reliable, production-grade AI systems — with a focus on retrieval quality, system reliability, and measurable operational impact.