At Hestabit, I built an end-to-end document AI pipeline that used OCR and an LLM to extract and enrich structured data. I designed AI-based validation and correction mechanisms that pushed OCR accuracy beyond 95% and cut manual review time by 95%.
I also shipped a Generative AI code-assessment platform with contextual retrieval and semantic analysis, auto-generating code reviews and deviation reports. In my projects, I built TalentSphere, an AI hiring platform with semantic candidate-job matching and voice interviews, and developed agentic research workflows with LangGraph and LangChain.
