My Work

Hakertrade (Under development)

A financial market dashboard with AI assistant features — live data, concurrency-limited fetch layer with retry/jitter, and an evolving AI-driven UI for market insights. Closed-source product.

  • React
  • TypeScript
  • Supabase
Visit Site →
Hakertrade

BG Remover

AI-powered image background remover with a model marketplace (U2-Net, ISNet, BiRefNet), remove.bg API integration, and alpha-matting for fine hair/fur edges.

  • React
  • FastAPI
  • rembg
  • HF Spaces
Visit Site → GitHub
BG Remover

Scoop

Your personal assistant for the overwhelming stream of school communications, event invitations, and family logistics. No more missed deadlines, forgotten permission slips, or double-booked schedules.

  • TanStack Start
  • Groq API
  • AI Agents
  • Vercel
Visit Site → GitHub
Scoop

llmScribe

Open-source Python tool that exports an entire codebase — directory tree plus full file contents — into one LLM-ready text file. Available on PyPI with CLI, GUI, and interactive terminal interfaces.

  • Python
  • Tkinter
  • CLI
  • PyPI
View on PyPI → GitHub
llmScribe

walliq

A wallpaper app with light/dark theming, favorites, and category filtering — built as both a web app and a full Flutter rewrite, backed by a lightweight GitHub-powered CDN.

  • Flutter
  • JavaScript
  • GitHub CDN
Visit Site → GitHub
walliq

NewsReels

A modern Indian news app built around an endless, TikTok-inspired scrolling experience for discovering and reading news across different categories.

  • React
  • TypeScript
  • RSS Feeds
Visit Site →
NewsReels

How I Build

I vibe code: I bring the architecture, product thinking, and Python; AI helps write the code. My job is knowing what to build, how it fits together, and how to ship it so it scales.

  • 1. Design the system

    Architecture first

    Services, data flow, database, and APIs are sketched before the first prompt, so the AI builds into a plan instead of guessing.

  • 2. Direct the AI

    Prompt with intent

    Clear context, constraints, and goals — and I review, question, and test everything the AI writes.

  • 3. Deploy & iterate

    Ship early, improve constantly

    Containerized, deployed, and monitored. Real users and real logs drive the next iteration.

Certifications

Explore All Certificates

Achievements

Highlights from platforms, building sessions, and recognition earned while shipping real software.

  • Published llmScribe on PyPI

    Python Package Index · 2026

    Released llmscribe as an installable open-source package — export any project into an LLM-ready text file via CLI, GUI, or terminal menu. Install with pip install llmscribe.

    View on PyPI →

  • Replit 10th Birthday — Top 8% of Builders

    Replit · May 2026

    Ranked in the top 8% of all builders. Shipped 2 apps in a single day with 64 AI agent interactions.

    🏅 Manager of Parallel Agent Recognized for high-output multitasking
    View Recap →

Tech Stacks

Python is my hands-on language. The rest is what I architect, deploy, and ship with AI assistance.

About Me

Amrito Kundu

I'm a self-taught builder and vibe coder. Python is the language I actually know, and I pair it with a solid grasp of system architecture, scalable deployment, and how AI fits into real products. I don't chase syntax — I design the system, direct AI to write the code, review it like an engineer, and ship it. That's how I've gone from Python tooling to React/TypeScript dashboards to Flutter apps.

Lately I've been deep in AI-assisted workflows — building tools like llmScribe that help AI models understand codebases faster, and products like Hakertrade and NewsReels that put useful technology directly in front of users. I like taking a messy idea and turning it into something structured, fast, and actually usable.