Divyakush Punjabi
Full Stack | AI/ML | Web Developer

Full Stack Rigor.
AI Awareness.
I am a Full Stack Developer and AI/ML Engineer with a proven track record of shipping production-grade applications that bridge the gap between complex backend logic and pixel-perfect frontend interfaces. My approach combines engineering rigor with a deep understanding of modern AI architectures.
Currently pursuing a B.Tech in Computer Engineering at VIT Vellore (GPA 8.17) with a Major in Artificial Intelligence from IIT Ropar, I have developed a unique perspective on building autonomous systems. From architecting transformer-based recommendation engines with 98.47% AUC-ROC to engineering SEO strategies that outrank multi-million dollar competitors, I thrive at the intersection of data and design.
Recognized as an AIR-140 finalist at IIT Bombay's National Entrepreneurship Challenge and a multi-time hackathon finalist (SIH 2025, EY Techathon 2025), I am dedicated to building software that is not only functional but transformative. With 20+ deployed projects, I am constantly pushing the boundaries of what's possible with React, Next.js, and Agentic AI.
- Sole technical lead for the product ecosystem, directing architecture and engineering execution across two live platforms while managing a student engineering team.
- Architected and developed Saturdays end-to-end — a consumer food-ordering marketplace owning discovery, ordering and payments.
- Built DineGuru, a restaurant operations platform handling inventory, kitchen ticketing and billing.
- Integrated both platforms over REST APIs and webhooks for real-time sync.
- Designed and built Research Atlas, an open-source AI research discovery platform unifying papers, repositories and governance resources into one semantic search experience.
- Implemented semantic retrieval with ChromaDB vector search over React/TypeScript and Python/FastAPI.
- Owned the end-to-end frontend for GovernAI Studio, an AI-governance training simulator, and contributed backend features within its Django, Celery and RAG pipeline.
- Shipped GovernAI's corporate website with an SEO strategy that improved organic visibility.
- Owned full lifecycle from architecture to deployment using React, Vite, TailwindCSS, and React Router.
- Engineered enterprise-grade SEO strategy that outranked multi-million dollar competitors in organic search.
- Delivered mobile-first, fully responsive UI with hero video and animated product sliders.
- Optimized WebP/WebM assets achieving 40% reduction in Time-to-Interactive.
- Created high-fidelity wireframes and interactive user flows in Figma.
- Translated prototypes into pixel-perfect responsive React components.
- Conducted user research and usability testing driving 20% increase in mobile retention.
- Led team Visionary Ventures in the National Entrepreneurship Challenge, representing VIT Vellore.
- Organized outreach campaigns and entrepreneurship-focused workshops.
- Oversaw and delegated multiple project tasks, ensuring timely and high-quality submissions.
Saturdays — Food Delivery Platform
Built and shipped a consumer food-ordering marketplace serving live traffic — menu discovery, cart and checkout flows with Stripe and PhonePe payment integration, on a React + MUI/Radix front end backed by a custom API.
DineGuru — Restaurant Management Platform
Multi-tenant restaurant operations backend: a FastAPI modular monolith over PostgreSQL with nine domain modules — identity & access, menu catalog, ingredient stock, orders & billing, recipe costing, procurement and analytics — with server-side tenancy scoping and a React dashboard.
GovernAI Research Atlas
Open-source AI research platform built on behalf of GovernAI, unifying papers, repositories and governance resources into one semantic search experience. Semantic retrieval runs on ChromaDB vector search with Sentence-Transformer embeddings across OpenAlex and GitHub, ranked by a custom relevance score.
GovernAI Studio — AI Governance Simulator
AI-governance training simulator delivering a persona-driven interface for governance and compliance scenarios. Owned the end-to-end frontend and extended core simulator logic in the backend's Django, Celery and hybrid RAG + LLM inference pipeline.
AI-Powered Content Recommendation Platform
Architected a transformer-based SASRec deep learning model trained on a 25M-parameter dataset. Deployed full-stack system with automated data ingestion pipelines.
AI & IoT Rockfall Prediction System
Engineered multi-sensor data fusion pipeline with real-time hazard detection ML model achieving 98% AUC-ROC — shortlisted for SIH 2025.
IoT Smart Home Automation System
Developed voice-activated smart home control system integrating Google Gemini API (LLM) for natural language command parsing. Ranked in top 12 out of 400+ teams.
Netra — Edge-AI Surveillance System
Autonomous person-tracking surveillance on constrained hardware: ESP32-CAM streams into a YOLOv8 detection pipeline driving pan-tilt servos over MQTT, with anomaly scoring and patrol heat-maps surfaced through a FastAPI + React console.
AlgoVerse — DSA Visualization Platform
Interactive algorithm and data-structure visualization platform, rendering sorting, graph and tree operations as animated 3D scenes to make execution and complexity legible step by step.
FPGA-based Smart Adaptive Traffic Light Controller
Implemented a real-time adaptive traffic management system on FPGA using Verilog. Optimized vehicle throughput by dynamically adjusting signal timing.
AI / ML
- Python, TensorFlow, PyTorch
- Scikit-learn, LangChain, LLMs
- NLP, Computer Vision
- SASRec, Predictive Modeling
Full Stack
- React.js, Next.js, TypeScript
- Node.js, Express.js, REST APIs
- FastAPI, Flask, SQL
- MongoDB, PostgreSQL
Frontend & UI
- Tailwind CSS, Radix UI, shadcn/ui
- Figma-to-Code, WCAG
- Zustand, Redux, Context API
- Responsive Design
DevOps & Tools
- Git, Docker, CI/CD
- Vite, Vercel, Vitest, Jest
- SEO Optimization
- Agile / Scrum