At Surekha Technologies, I found an internal workflow automation opportunity and built an end-to-end solution using Python and LLM-based workflows. I also designed MCP servers that exposed internal tools for agents to use in multi-step workflows.
I supported production deployment and monitoring of an LLM-based product, investigating inaccurate outputs and comparing approaches to improve reliability. I built FastAPI REST APIs and React applications, including an OCR document extraction pipeline and a YOLOv11 monitoring system.
At Elevate Labs, I built a reproducible machine learning pipeline on 40,000+ labeled articles that achieved 93%+ accuracy, and reduced RMSE by 18% on a regression task. My projects include an appointment assistant that uses deterministic backend logic to validate scheduling conflicts, along with agent workflows and RAG systems.
