Full-stack engineer. Shipping systems that run themselves.

Challenge

Build a complete nutraceutical storefront a small wellness brand could trust with real transactions - multi-variant products, secure checkout, live inventory, and an admin dashboard a non-technical operator could run the business from.

Solution

Next.js SSR/SSG frontend on Vercel, Node.js + Express REST API, PostgreSQL via Neon, Razorpay payment flows, and Sentry.io production monitoring - end-to-end commerce architecture with observability baked in from day one.

Highlights

• Production storefront: browsing, secure checkout, customer accounts
• Razorpay integration with automated order processing
• RESTful APIs for products, variants, categories, customers, orders
• SEO-optimized SSR/SSG with category filtering, dynamic pricing
• Admin dashboard for product, variant, inventory & order management

Challenge

Build a personal finance system that doesn't just record - it watches, warns, and reports. Hardened auth, real-time alerts, and AI-driven intelligence on top of raw transactions.

Solution

React 18 + Node.js + MongoDB with Socket.io for real-time events and n8n AI agents orchestrating five automated workflows - from notifications to anomaly detection across spending patterns.

Highlights

• Real-time budget alerts via Socket.io
• 26 REST endpoints: JWT access + refresh rotation, Bcrypt, Helmet, Redis rate-limit
• 5 n8n AI workflows: notifications, budget-breach, weekly AI summaries, monthly HTML reports, anomaly detection
• 4 MongoDB schemas · compound indexes · aggregation pipelines
• Promise.all parallel execution for dashboard performance

Challenge

Traditional fixed-interval traffic signals cause inefficiency at intersections with variable lane densities. Build a system that dynamically adjusts signal timing based on real-time vehicle counts per lane.

Solution

Custom-trained YOLOv5 model on a curated vehicle dataset, a lane-density adaptive timing algorithm, and a Python + FastAPI real-time camera-to-signal pipeline - closing the loop from perception to control.

Highlights

• YOLOv5 trained on custom dataset for per-lane vehicle counting
• Lane-density adaptive timing algorithm with dynamic signal adjustment
• Python + FastAPI real-time inference pipeline
• OpenCV integration for frame capture & preprocessing
• Final-year B.E. major research project

Challenge

Build a feature-complete email client with automated intelligence - not just CRUD, but label inference, notifications, and threading that actually works.

Solution

Full MERN stack email client with complete CRUD operations and n8n workflow automation layered on top for auto-labeling and event-driven notifications.

Highlights

• Complete CRUD: inbox, compose, labels, threads, search
• n8n workflows for auto-labeling & notifications
• Full-text search, read/unread states, label filtering
• MongoDB + Express + React full stack