I’m an End-to-End Computer Vision Pipeline Engineer specializing in building production-grade AI systems that go from raw CCTV footage to real-time actionable insights.
I design, train, optimize, and deploy complete CV pipelines (detection → tracking → real-time backend → dashboard) that are currently running live in manufacturing environments.
Currently building production CV systems at CoolR Group (Remote AI/Software Engineering Intern).
Computer Vision
YOLO-seg • SAM • DINO • BoT-SORT • OpenCV • Multi-object Tracking
AI & LLM
OpenAI • LangChain • MCP • RAG • Hugging Face • PyTorch • TanStack AI • Zod
Backend & Real-time Systems
Node.js • Express • Redis (Pub/Sub) • WebSockets • SSE • PostgreSQL • MongoDB
Infrastructure
Docker • Modal (Serverless GPU) • AWS (ECS) • GitHub Actions • CI/CD
Core Expertise
End-to-End CV Pipelines • Human-in-the-Loop Workflows • Multi-step Agentic Systems • Low-latency Streaming
SackEye — Industrial Inventory Monitoring System
End-to-end Computer Vision + Real-time Backend Pipeline
→ Live Demo (Video + Detection Gallery): https://sackeye.duckdns.org/public/job/69d11e73eab4aecc86878cb1?group=777
Key Results
- 95% detection accuracy with YOLO-seg + SAM + BoT-SORT
- Reduced manual auditing from ~3 hours → under 5 minutes per session
- Near 100% operational accuracy via Human-in-the-Loop verification
- 80% lower infrastructure cost using Modal serverless GPU
- Live in production across 5+ manufacturing units
Repositories
-
CoolR Group (Oct 2025 – Present)
AI/Software Engineering Intern — Built streaming LLM chatbot with SSE + MCP tool orchestration (25+ enterprise APIs) and real-time CV pipelines. -
Zenthos (Aug–Sep 2025)
Software Engineering Intern — Developed MelaPass, a secure offline ticketing system with RSA signatures and distributed sync engine. -
Upwork (2025)
Freelance ML Engineer — Delivered YOLO-based detection systems and Azure ML event-driven retraining pipelines (5.0/5.0 rating).
- Codeforces: Specialist (Max Rating 1475)
- LeetCode: 600+ problems solved (350+ Medium, 110+ Hard)
- Built GPT-2 transformer from scratch in PyTorch and instruction-tuned it to 97.5% accuracy
B.Tech in Computer Science and Engineering
Maharaja Agrasen Institute of Technology (MAIT), New Delhi
2022 – 2026