AI engineer & software developer building agentic systems - from context engines to production backends along with evaluation.
X • LinkedIn • Email • Portfolio
- AI Engineer @ EpicWise — building the core context engine: LangGraph pipelines, a public MCP server, and hybrid Pinecone + Neo4j retrieval.
- Backend Developer Intern @ FaxLab — shipped course catalog, auth, and dashboards for an AI course platform, with zero QA rollbacks.
- Evident — turns any URL into structured, typed data with transparent confidence scores for AI agents.
- RepoGraph AI — VS Code extension that turns mapped as a live dependency graph and answered with file:line proof. And your AI agent gets a one-file brief that cuts its context tokens by up to 40%. Right inside VS Code — install once, use it in every project.
- Buildify Labs — LangGraph (ReAct) agent that converts natural language into SQL and live dashboards.
- Lips Sync Face Tracker - Keyframe-based face tracking annotation tool for lip-sync: draw, move & resize speaker bounding boxes with forward-fill (no interpolation), mark absent segments, and export per-frame pickle.
- InvoiceGuard-AI — ai-powered invoice and Purchase Order (PO) mismatch detection tool inspired by the finance automation workflows offered by peakflo
- GRS Worker Portal — production worker-management system for attendance, payroll, and admin operations.
- Attendance Tracker — real-time PWA attendance system with role-based access control.
- Postly — real-time social platform using Socket.IO with Kafka-based async event processing.
- AI Context Engine — RAG + vector search backend that turns documents into structured insights.
- TwitterBotAI — AI-powered tweet automation using Gemini, pulling from news and RSS sources.
- Contributor to Joplin — pull requests merged into the desktop app and sync engine by upstream maintainers.
Currently exploring agent reliability & evals, fine-tuning small models, and graph-based retrieval at scale.
Also working with: LangGraph · MCP · RAG · Pinecone · DeepEval · Fine-tuning · LLM Evals