M↗ ML ROADGitHub ↗
AN OPEN LEARNING ROADMAP

From first model
to real AI systems.

One clear path through machine learning, deep learning, LLMs, retrieval and agents. Learn the ideas, use the resources, then build something that works.

Explore the roadmap ↗View source ↗
CURATED BY ILGIZ SATKYNOV · BUILT FOR CURIOUS BUILDERS
01 / THE ROADMAP

Pick a stage.
Build your way forward.

Each stage gives you a foundation for the next. The projects turn study into practice.

01
⌘

Foundations

Python, data tools, linear algebra, calculus, probability and statistics.

PYTHON · NUMPY · MATH
02
◇

Machine learning

Regression, classification, trees, clustering, evaluation and sound experiments.

SCIKIT-LEARN · DATA
03
◎

Deep learning

Neural networks, backpropagation, computer vision, NLP and transformers.

PYTORCH · TRANSFORMERS
04
✳

LLM engineering

Prompting, embeddings, retrieval, RAG, structured outputs and evaluation.

LLMS · RETRIEVAL · EVALS
05
⤳

Agentic AI

Tool use, agent loops, state, MCP, human oversight and multi-agent patterns.

TOOLS · MCP · AGENTS
06
▣

Production systems

APIs, deployment, monitoring, security, cost and reliability in the real world.

APIS · CLOUD · OBSERVABILITY
02 / START LEARNING

Good resources.
A practical order.

01
Machine Learning Crash CourseGOOGLE · MACHINE LEARNING
↗02
Deep Learning SpecializationDEEPLEARNING.AI · NEURAL NETWORKS
↗03
CS224N: NLP with Deep LearningSTANFORD · NLP
↗04
The LLM CourseHUGGING FACE · TRANSFORMERS
↗05
Building Effective AgentsANTHROPIC · AGENT DESIGN
↗
Explore the full resource collection on GitHub ↗
03 / PROJECT IDEAS

Build to understand.

Suggested exercises, from a first implementation to a complete AI application. These are ideas, not claims of completed work.

START HERE / 01

Linear regression
from scratch

Implement gradient descent, plot the fit and compare against a library model.

ML FOUNDATIONS ↗
GO DEEPER / 02

Document Q&A
with RAG

Index documents, retrieve relevant passages and measure answer quality.

LLM ENGINEERING ↗
BUILD A SYSTEM / 03

Tool-using
research agent

Connect search and reading tools, keep evidence and test against real tasks.

AGENTIC AI ↗
ABOUT THE PROJECT

A road worth
building on.

ML Road is a learning resource curated by Ilgiz Satkynov. It connects fundamentals with modern AI engineering through courses, papers and hands-on ideas.

The resource collection began with the open-source ml-road project. Original authors and resource creators retain credit for their work.

Explore on GitHub ↗