Chand Ansari

• Building and shipping production e-commerce websites for clients — product catalogues, cart and checkout flows, and order management — using Next.js, Node.js and MongoDB/PostgreSQL.
• Developing LMS platforms with live class capabilities — video lesson delivery, class scheduling and student dashboards — for online education clients.
• Owning features end-to-end across the stack: REST API design, authentication and role-based access control, and deployment to production.
• Shipped additional client work — company websites with blogs, landing pages and utility tools such as a browser-based PDF editor — with SEO-ready structure, fully responsive layouts and fast load times.

• Architected event-driven backend services using Apache Kafka for asynchronous order processing and Redis for high-throughput caching, reducing database load by 40% and improving API latency.
• Containerised full-stack applications with Docker and automated zero-downtime deployments via GitHub Actions CI/CD pipelines.
• Designed and built RESTful APIs with strict input validation, pagination and JWT-based RBAC middleware using Node.js, Express and Next.js.
• Deployed production applications on AWS EC2 with an Nginx reverse proxy, SSL termination via Let's Encrypt and PM2 process management.

• Built the full frontend and backend of a RAG-powered conversational AI with a ChatGPT-style interface, streaming responses and persistent chat history.
• Implemented end-to-end Firebase OAuth for Google and GitHub with MongoDB as the user store, handling token exchange and secure profile persistence.
• Owned the chat history module: designed MongoDB schemas, built paginated history APIs and managed session-based context windows to optimise LLM token usage.
• Collaborated in a cross-functional agile team to design API contracts, conduct code reviews and ship features faster.

Next.js 16 · React 19 · TypeScript · Supabase · PostgreSQL · Row-Level Security · Server Actions · Tailwind CSS 4 · Vercel

• Shipped three role-gated portals behind one database, resolving roles at the edge so unauthorised users never reach a protected route.
• Made Row-Level Security the authorization boundary across roughly 45 policies and 26 tables, so a leaked client key exposes no tenant data.
• Built a reducing-balance EMI engine with drift-corrected amortisation schedules, dual-mode prepayment modelling and grace-period-aware late fees, mirrored exactly in a SQL function.

Next.js 16 · React 19 · TypeScript · PostgreSQL · MongoDB · Redis · Socket.IO · Razorpay · Playwright · Docker

• Ran a polyglot persistence layer: PostgreSQL for relational course data, MongoDB for reviews, AI conversations and hash-sharded activity logs, and Redis for caching, rate limits and the job queue.
• Enforced role-based access twice — edge route guarding in proxy.ts, plus cryptographic JWT verification on every API route with roles read from server-controlled app metadata.
• Built HMAC-SHA256 verified Razorpay webhooks behind Redis idempotency locks, so replayed payment events are detected and double-charging is impossible.

Next.js 14 · React 18 · MongoDB Atlas · Mongoose · Cloudinary · NextAuth.js · Tailwind CSS

• Streamed uploads straight to Cloudinary with real XHR progress and cancellable transfers, then generated thumbnails server-side from a URL transform — no temp files written to disk.
• Modelled the domain in four Mongoose schemas with ten strategic indexes, including a compound text index powering search and a unique compound index guaranteeing one watch-history row per user and video pair.
• Implemented threaded comments through a self-referencing parent reference, hydrating replies in parallel so latency stays flat as thread depth grows, with cascading deletion.

Next.js 16 · Supabase · PostgreSQL · pgvector · Groq · Razorpay · Tailwind CSS

• Architected a production electronics storefront on Next.js 16 and Supabase with a faceted catalogue and Razorpay plus COD checkout, computing all pricing server-side in Postgres so a tampered client cannot set its own totals.
• Designed a RAG shopping assistant on local bge-small embeddings — no external model API, no per-request cost, no data leaving the machine — using pgvector hybrid search that fuses vector and full-text results via Reciprocal Rank Fusion, with answers generated by Groq.
• Solved personal-data isolation by never embedding customer orders: they are fetched per request and scoped to the session user, making cross-customer leakage structurally impossible rather than merely filtered.

React Native · Expo SDK 57 · TypeScript · React Navigation · AsyncStorage · EAS Build

• Built the complete booking funnel across twelve screens and ten routes: browse, service detail, slot and worker selection, checkout, confirmation, live tracking, history and ratings.
• Encoded a six-state booking lifecycle with timeline, per-status design tokens, cancellation rules and rating eligibility as domain invariants in a single reducer.
• Shipped a wallet with top-ups, a credit and debit ledger and rewards, plus a coupon engine handling flat and percentage discounts with minimum-order gates and caps.