Arnab Dey

I’m an AI Engineer and Agentic Systems specialist with 2.5+ years of experience building LLM-driven, event-driven automation pipelines, RAG retrieval systems, and MCP-integrated workflows. I focus on context engineering, structured outputs, and reliable tool-based agent execution.

At Unisys, I designed and integrated MCP workflows across 5+ enterprise systems to enable entitlement-aware context construction for production LLM workflows. I built RAG pipelines over thousands of operational records using FAISS vector stores and Azure OpenAI embeddings, improving contextual retrieval quality for production decisioning.

I also engineered event-driven automation in Python and designed RBAC/policy-based access control logic aligned with governance and audit needs. By automating server lifecycle, data lineage tracking, and operational health monitoring with Azure Logic Apps and Hybrid Workers, I reduced MTTR by ~40% and eliminated manual effort for 60%+ of recurring incident types.