santhos kumar

I've built FixFlow, an agentic AI system that classifies software bugs and automates the Jira ticket lifecycle. Using n8n, OpenAI/Gemini APIs, Supabase pgvector, and FastAPI, I delivered 92.8% classification accuracy and reduced known-incident MTTR from 3.0 hours to 2.0 seconds.

My work spans multi-agent pipelines, RAG systems, vector search, embeddings, and LLM-powered workflows built with Python, LangChain, LangGraph, and Azure. I design agent capabilities that can be exposed through REST APIs and integrated into external systems in real time.

At Fiber Sense, I built an ML-powered Android application for real-time cotton classification and developed Azure-integrated prediction workflows for scalable inference, monitoring, and deployment. As an SDE Intern at Bharat FIH, I modernized the DL Recruit hiring portal with AngularJS and TypeScript, integrated REST APIs, and contributed through Agile code review and testing practices.

I've also developed reinforcement-learning edge-cloud IoMT systems, OCR-based Android GST filing tools, interactive analytics dashboards, and published research in deep learning and edge-cloud coordination. My work has led to a VIT SEED research grant and multiple patent applications.