MachineLearningRoadmap is a comprehensive, hands-on learning program for becoming a machine learning engineer. It guides learners through building neural networks from the most basic concepts all the way to modern architectures like transformers and diffusion models. The course emphasizes writing code yourself rather than just using pre-made tools, and includes three portfolio projects that students can showcase to potential employers. Written in Russian, it spans from foundational math to production-ready deployments, with a philosophy of 80% practice and 20% theory.
How It Works
You find a structured roadmap that promises to teach you neural networks the right way — with code that actually runs.
You start with the math module, learning derivatives, matrices, and probability basics so you truly understand what you're doing.
Step by step, you create perceptrons, MLPs, CNNs, and transformers — writing every piece yourself, not just calling pre-made tools.
Explore image models like Vision Transformers, CLIP, and image generation with diffusion models
Learn RLHF, instruction tuning, and how to make language models behave the way you want
Discover how to work with graph neural networks and models that learn without labels
You complete three capstone projects — an image classifier, your own mini language model, and a production-ready assistant.
With three real projects on GitHub and deep understanding of how neural networks work, you ace your interviews and get hired.
Star Growth
Repurpose is a Pro feature
Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.
Unlock RepurposeSimilar repos coming soon.