A practical AI/ML engineer path — Python and mathematics through machine learning, deep learning, LLM applications, agents, and production deployment. 14 modules, one arc, server-graded code and real widgets from lesson one — not video.
Every module builds on the last. Modules 1–4 are the free foundation — genuinely usable on their own, not a trial that expires. Module 5 onward is where the "AI/ML engineer" part of the title actually gets earned.
Python from variables through NumPy, pandas, and Git — the tools every later module assumes you already have.
How data structures and algorithms actually work, plus the math (vectors, matrices, probability, distributions) every ML algorithm is built on.
SQL for real datasets, the calculus/statistics behind training, and the classical ML workflow — regression, trees, boosting, evaluation.
Backpropagation, training loops, CNNs, RNNs/LSTMs, attention, and transformers — from first principles, not a black-box library call.
Prompt engineering, embeddings and vector search, retrieval-augmented generation, fine-tuning, and building autonomous agents that use tools.
Experiment tracking, CI/CD, model serving, responsible AI, and system design — plus a focused introduction to RL as its own learning paradigm.
Documenting and presenting real engineering work, then a final integrated project that ties the whole arc together.
Roughly, as of when this was written — subscription pricing shifts, so treat these as ballpark, not a guarantee:
| Platform | Typical monthly price |
|---|---|
| epochai Pro | $9.99/mo (or $149 one-time, Lifetime) |
| DataCamp Premium | ~$25–39/mo |
| Codecademy Pro | ~$20–40/mo |
epochai's free tier (Modules 1–4) has no time limit and no card required — most competitor free tiers are a single course or a short trial.
No signup, no video — a real interactive widget from the actual curriculum.
Modules 1–4 — Programming Fundamentals through Linear Algebra & Stats — no card required.