I'm a full-stack developer focused on TypeScript, Node.js, Next.js, and backend development, with hands-on experience building APIs, authentication systems, payment flows, Dockerized applications, and AI-powered applications.
- Frontend: Next.js, React, TypeScript, Tailwind CSS, shadcn/ui, Tanstack form, Tanstack Query, Tanstack Router
- Backend: Node.js, Express.js, MongoDB, Mongoose, JWT Authentication, OAuth
- Database: MongoDB, PostgreSQL, Redis
- DevOps: Docker, Nginx reverse proxy, Kubernetes basics, AWS EC2/EKS practice
- Tools: Git, GitHub, Postman, VS Code, intellij, apidog
A Retrieval-Augmented Generation (RAG) application built from scratch. It processes PDF, Markdown, and text documents, generates embeddings, stores them in PostgreSQL using pgvector, and retrieves relevant document chunks using vector similarity search. The retrieved context is then provided to an LLM to generate grounded answers with source references.
Tech used: TypeScript, Node.js, Fastify, PostgreSQL, pgvector, Groq, Zod, Docker
An AI-powered document extraction API that converts unstructured invoices and resumes into validated structured JSON. It supports PDF, TXT, and Markdown files, extracts document text, sends it to an LLM using Groq and the AI SDK, and validates the generated output using Zod schemas. The invoice workflow also includes deterministic business-rule checks to detect missing critical fields and flag documents that require human review.
Tech used: TypeScript, Node.js, Fastify, Groq, Vercel AI SDK, Zod, pdf-parse
A local Windows print agent for web-based ERP/POS systems that bridges browser applications with thermal printers. It supports Windows and network ESC/POS printing, printer configuration through a local dashboard, SQLite-backed settings, failed print-job handling and retries, Windows auto-start, and standalone EXE packaging.
Tech used: Node.js, TypeScript, Express.js, SQLite, ESC/POS, Pug, Zod, esbuild, Windows
A full-stack restaurant ordering application with authentication, authorization, payment integration, and admin/user features. The idempotency model is used to prevent duplicate processing of the same request, especially during payment or order creation.
Tech used: Next.js, TypeScript, Node.js, Express, MongoDB, JWT, OAuth, eSewa
A secure backend authentication API built with Node.js, Express, TypeScript, MongoDB, and JWT. It includes user registration, login, refresh tokens, protected routes, Google OAuth, email verification, password reset, admin role protection, and 2FA/TOTP.
Tech used: Node.js, Express.js, TypeScript, MongoDB, Mongoose, JWT, Cookies, Google OAuth, Nodemailer, Zod, 2FA/TOTP
A practice project showing reverse proxy, multiple services, and Docker Compose networking.
A local Kubernetes deployment practice project with deployment files, services, ingress, and TLS learning.
A full-stack real estate web application built with Node.js, Express, MongoDB, Mongoose, and Pug. It includes property listings, user authentication, JWT cookies, protected routes, image upload, image resizing, pagination, and password reset.
Tech used: Node.js, Express.js, MongoDB, Mongoose, Pug, JWT, Cookies, Multer, Sharp, Nodemailer
A modern e-commerce application built with Next.js, React, Supabase, NextAuth, Zustand, and Tailwind CSS. It includes product listing, cart management, protected profile/checkout pages, checkout session creation, and eSewa payment success flow.
Tech used: Next.js, React, JavaScript, Supabase, NextAuth, Zustand, Tailwind CSS, eSewa, Vercel
- GitHub: sunflowerIU
- LinkedIn: linkedin.com/in/amit-thing-650870306