At TAG Ecosystem, I’m developing the backend and RAG layer for SentinelAI, a supply-chain disruption agent that flags shipments at risk of arriving late and prepares evidence-grounded briefings. I designed an evidence trail that checks claims against tool results and blocks briefings that cite evidence that doesn’t exist.
For Vault AI, I worked on an offline document Q&A assistant using Ollama, ChromaDB, and FastAPI. I owned its evaluation, creating a benchmark with realistic documents and negative cases; the system reached 92% accuracy.
I also built document ingestion for TalentMatch AI, parsing PDF, DOCX, and TXT files with page-level source tracking so claims can be traced to the resume. Across these projects, I focus on testing AI systems rather than assuming they work.
Before my AI engineering work, I shipped mobile and full-stack apps. At Synergyy.io, I built a Flutter operations app that improved task-tracking efficiency by 40% and cut load time by 30%; at E86.NG, I designed backend architecture and REST APIs for VIBETEST and a school management system.
