Deepesh Yadav
I build AI-native products end to end — from RAG pipelines to production Next.js apps.
Selected Work
Veda_AI
AI-powered exam generator for teachers. Monorepo (Next.js, Express, Socket.io) with async, non-blocking question generation via GPT-4o-mini, real-time WebSocket status updates, and one-click PDF export.
Cortex
Real-time collaboration tool with Socket.io, containerized via Docker Compose, deployed through a GitHub Actions CI/CD pipeline.
Camera-Event-Ingestion
High-volume event ingestion service in Go. REST API ingests camera/telemetry events via POST endpoints, with a concurrent worker-pool (goroutines + channels) decoupling ingestion throughput from write latency to MongoDB. /health and /metrics endpoints for uptime observability.
AlgoCoach-AI
Multi-agent competitive programming assistant. LangGraph-orchestrated agents with RAG pipelines for context-aware DSA hints and code review.
Hardware Trojan Detection
Co-authored ML research on hardware trojan detection in IC design. SVM-RBF classifier, 99.54% cross-validation accuracy across an 11-class classification task.
PagePilot
RAG-powered Chrome extension for in-page Q&A and summarization — built on Chrome MV3, running a retrieval pipeline against live page content.
Portfolio
This site — a cinematic dark portfolio built on TanStack Start + Framer Motion magic-move morphing between Hero and About states, backed by a sky-backdrop parallax.
More shipping soon —follow on GitHub
The Toolkit
Languages
AI / ML
Frontend
Backend & Data
Tools
Beyond Code
Open Source
Shipping PRs, writing docs, and treating open source as a first-class part of my craft — not a side quest.
Basketball
Court time keeps the reps sharp — teamwork, spacing, and knowing when to pass the ball. Same rules apply in code.
Chess
Positional thinking, patience, and the discipline of thinking three moves ahead. Debugging with a clock.
Let's build something.
Open to internships, AI/full-stack roles, and open-source collabs. The inbox is faster than the DMs.
deepeshiiitbhopal@gmail.com