Jatin Mahajan

Full-Stack Architect • Vision AI Engineer • Systems Builder

---

👋 About Me

I’m an engineer dedicated to building intelligent, scalable, high-performance software systems.
My work spans across full-stack development, computer vision, LLM-based automation, distributed backend systems, and developer tooling.

I thrive on solving challenging engineering problems and taking products from idea → architecture → execution → production.

Currently, I’m co-building Quantivision — a next-gen Vision AI and Agentic Automation company focused on real-time intelligence, autonomous analytics, and high-impact engineering for modern enterprises.

---

🚀 Core Engineering Stack

Frontend Engineering

• React • Next.js 15 • React 19 • TypeScript
• Tailwind CSS • Three.js • React Three Fiber • Framer Motion
• Building immersive experiences and interactive 3D product sites

Backend Engineering

• Golang (Gin, gRPC, sqlc, concurrency patterns)
• Node.js / Express
• Python (FastAPI, Flask)

Systems, Distributed & Real-Time Engineering

• Kafka (Confluent stack, kafka-go)
• Custom logging pipelines with Kafka producers/consumers
• File-backed timestamp-triggered database in Go
• Pub/Sub systems with at-least-once delivery
• Microservice orchestration & modular codebases

AI / Machine Learning

• Vision AI (SAM, pose models, object tracking, sports analytics)
• Model fine-tuning (e.g., Med-GEMMA-4B-IT, Med-GEMMA-3)
• PyTorch training pipelines
• Attention heatmaps & CV interpretability
• Image/video compression & preprocessing (OpenCV, FFmpeg)

DevOps & Cloud

• Docker • Kubernetes • Helm
• AWS • GCP
• CI/CD workflows
• Nginx • PM2

Databases

• PostgreSQL • MySQL • MongoDB
• Redis
• sqlc (type-safe Go queries)
• golang-migrate/migrate

Tools & Creative Tech

• VS Code • Figma • Postman
• Blender • Unity (for experiments)

---

🛠️ Notable Work & Projects

🔵 Quantivision — Vision AI + Agentic Intelligence

• Real-time sports analytics (player tracking, event understanding)
• Multi-model perception pipelines
• Agentic LLM workflows for automation
• Enterprise-scale dashboards and intelligence systems

⚙️ Timestamp-Triggered Database (Go)

• Custom file-backed DB
• gRPC-based subscription system
• At-least-once delivery pub/sub
• Pluggable scheduling engine

🧪 Medical AI Pipeline

• Fine-tuning Med-GEMMA-3 & 4B-IT on medical datasets
• Pneumothorax detection with attention heatmaps
• Torch training loops with configurable backbones

🟧 Rust CLI Task Tracker

• Modular architecture (task.rs, db.rs, etc.)
• Due dates, priorities, categories
• Optional SQLite/sled backend
• Terminal UI vision using Ratatui

🌌 Portfolio Website (React + Three.js)

• Galaxy/universe themed
• Physics-based interactions using Cannon.js
• Immersive 3D storytelling

---

🧠 Current Focus

• Building real-time multi-model Vision AI systems
• Scaling LLM agent pipelines at Quantivision
• Mastering Go for high-performance backend architecture
• Training, optimizing, and visualizing ML models
• Refining Rust skills for systems programming
• Designing developer-friendly toolchains and CLIs

---

📊 GitHub Stats

---

✨ Thanks for visiting!