At a stealth AI startup, I architected and implemented a production agentic runtime for multi-step autonomous workflows, including planning, tool execution, replanning, approval handling, and response synthesis.
I built scalable AI backend services and enterprise RAG pipelines, and integrated tools through MCP and FastMCP with execution controls and approval gates. I also worked with multiple LLM providers, including Azure OpenAI and AWS Bedrock.
Previously, at AI Dev Lab, I built transformer-based NLP and LLM applications and RAG workflows. At Scribe Audio, I developed machine learning and NLP pipelines for speech analytics, transcription, and audio intelligence.

