Stack
The tools I build with
Everything I reach for, from TypeScript and Postgres in production to Python and TensorFlow for models, and ARM assembly on a microcontroller.
Languages
What I write in, from TypeScript on the web to C and assembly on a microcontroller.
- TypeScript
- JavaScript
- Python
- Java
- Kotlin
- C++
- C
- SQL
- ARM Assembly
Frontend
What people touch.
- React
- Next.js
- Tailwind CSS
- HTML5
- CSS
Backend
Where the rules live.
- Node.js
- Express
- REST APIs
- Socket.io
- Prisma
- Telegram Bot API
Databases
The part you can't lose.
- PostgreSQL
- MySQL
- MongoDB
- Redis
- Supabase
Mobile
Native Android, written in Kotlin.
- Kotlin
- Android
Machine learning & deep learning
Classical ML to neural networks.
- Python
- TensorFlow
- scikit-learn
Infrastructure & tooling
Keeping it shipped, versioned and awake at 3am.
- Docker
- Nginx
- Linux
- Git
- GitHub
- CI/CD
- Sentry
Payments
Money, handled carefully.
- bKash
- Cash on delivery
Embedded systems & IoT
Firmware on real hardware, and RFID attendance in production.
- ARM Assembly
- C
- STM32
- Keil µVision
- IoT / RFID
AI-assisted engineering
I build with AI agents, not just alongside them
Coding agents and the harnesses around them are part of how I ship. I use them with review, tests and judgement, so the speed doesn't cost correctness.
Claude Code
Anthropic's agentic coding CLI, for planning changes, multi-file refactors, reviews and test runs straight from the terminal.
Codex
OpenAI's coding agent, for delegating well-scoped tasks and for generating and reviewing code alongside my own work.
AI agent harnesses
Wiring models into real workflows: agent loops, tool calling and the context setup that makes AI output reliable, not just impressive.
Data scraping
Collecting and cleaning web data with Python to build datasets for models, research and automation.
See it in use
Tools are only as good as what they ship
The projects show this stack in production; the coursework shows the foundations under it.