Guna Sekhar

I've built an LLM-assisted document processing platform that converts unstructured PDFs and images into validated structured data. The system combines OCR, FastAPI, LLM extraction, Pydantic models, JSON Schema, PostgreSQL, deterministic validation, and human review.

At Micro1, I developed Python programming tasks, technical prompts, evaluation criteria, and test cases for Large Language Models. I evaluated and debugged generated Python solutions, analyzed failure modes, and contributed structured feedback to RLHF-style evaluation workflows.

I'm currently supporting ITREEN DATA AND SYSTEMS by provisioning Azure infrastructure, configuring Windows and Linux services, managing access controls, and resolving production incidents. I also work with clients to turn evolving operational needs into scalable infrastructure updates.

Across projects, I've built retrieval and recommendation systems using TF-IDF and cosine similarity, plus a full-stack workflow platform with Node.js, Express.js, MongoDB, JWT authentication, and automated testing. I focus on making AI and cloud systems testable, reliable, and safe for downstream use.