# Crackr

> Learn ML engineering by building real systems from scratch.

Crackr is a project-based learning platform for machine-learning engineers. Learners build systems such as tokenizers, transformers, vector databases, RAG pipelines, autograd engines, and diffusion models in their own repositories. Each project is divided into focused tasks with technical explanations, implementation guidance, and automated evaluations.

## How It Works

1. Choose an ML system to understand.
2. Build it locally using your own editor, terminal, and tools.
3. Push the implementation and run automated evaluations against it.
4. Fix failures and continue through the project one task at a time.

## Projects

- Build a Tokenizer from Scratch
- Build a Transformer from Scratch
- Build a Vector Database from Scratch
- Build a RAG System from Scratch
- Build an Autograd Engine from Scratch
- Build a Diffusion Model from Scratch

## Key Pages

- [Home](https://crackr.dev/): Product overview and project catalog.
- [Sign in](https://app.crackr.dev/sign-in): Create an account or continue building.
- [Privacy Policy](https://crackr.dev/privacy): How data is collected and used.
- [Terms of Service](https://crackr.dev/terms): Platform terms and conditions.
- [Refund Policy](https://crackr.dev/refund): Refund terms for paid access.

## Contact

- Email: hi@crackr.dev
- Website: https://crackr.dev

Last updated: 2026-08-22