Skip to content
Epoch

The AI engineering roadmap

From first line of code to AI engineer.

A semester-by-semester path through machine learning, deep learning and generative AI. Hand-picked free resources, real projects and progress you can see. Built for engineering students.

Start the roadmapExplore career tracks
  • 7epochs
  • 56topics
  • 35free resources
  • 7portfolio projects

Why it is called Epoch

In machine learning, an epoch is one complete pass through the data. Here, it is one complete stage of your growth. Seven of them take you from your first line of Python to shipping generative AI that people actually use.

Mapped to your degree

Every epoch lines up with a semester, so learning AI runs alongside your coursework instead of against it.

Hand-picked and free

Each epoch points to the best free courses, books and papers we could find. No paywalls, no filler.

Built on real projects

Every epoch ends with a portfolio project, so you graduate with proof, not just certificates.

At a glance

Everything you need. Nothing you don't.

Seven epochs, each with a clear outcome, a checklist of topics, the best free resources and a project that proves what you learned.

The big picture

AI, ML and Gen AI, finally untangled.

Artificial intelligence is the goal. Machine learning is how most of it works today. Deep learning scaled it up, and generative AI is its newest layer. You will learn all four, in that order.

Generative AI sits inside deep learning, which sits inside machine learning, which sits inside artificial intelligence.

7

Epochs

From Python basics to production Gen AI, mapped across eight semesters.

56

Topics to master

Each one a clear, checkable milestone with a one-line summary of what it covers.

35

Free resources

Courses, books and papers from Stanford, MIT, Harvard, Google and the builders themselves.

7

Portfolio projects

Build something real at the end of every epoch.

Progress that sticks.

Tick off topics as you learn. Your progress saves privately in this browser, with no account and no sign-up.

0%

0 of 56 topics done

Start tracking

The roadmap

Seven epochs.
One AI engineer.

Work through each epoch in order. Tick off topics as you learn them, build the project, then move on. Durations assume six to eight hours a week alongside your coursework.

0%

Your progress

0 of 56 topics complete

Saved privately in this browser. No account needed.

Epoch 01Year 1 · Semesters 1–216 weeks

Foundations

Code, and the math behind it.

0/8

Topics complete

Learn to think in code and in vectors. Every model you will ever train is written in these two languages, so this is where the real advantage is built.

By the end of this epoch

You write clean Python, explore any dataset with confidence and follow the math behind a model.

Topics

0 of 8 done

Tools

  • Python
  • Jupyter
  • NumPy
  • pandas
  • Matplotlib
  • Git

Epoch project

A data story of your own

Pick a public dataset you genuinely care about. Clean it, explore it and publish a notebook that answers three sharp questions with clear charts.

The toolkit you will master along the way

  • Python
  • NumPy
  • pandas
  • scikit-learn
  • XGBoost
  • PyTorch
  • Hugging Face
  • Transformers
  • LangChain
  • LlamaIndex
  • FAISS
  • Diffusers
  • FastAPI
  • Docker
  • MLflow
  • vLLM
  • Kaggle
  • Google Colab

Career tracks

Where this can take you.

Every track starts from the same foundations. Your interests decide which epochs you go deepest on.

  • Machine Learning Engineer

    Designs, trains and deploys the models inside real products.

    You own the path from raw data to a model serving predictions at scale, balancing accuracy, speed and cost.

    • Python
    • PyTorch
    • MLOps
    • System design

    Go deepest on

    Epoch 02Epoch 03Epoch 06

  • Generative AI Engineer

    Builds copilots, agents and retrieval systems on top of LLMs.

    You turn foundation models into dependable products, with prompts, retrieval, tools and evaluations that hold up.

    • LLMs
    • RAG
    • Agents
    • Evaluation

    Go deepest on

    Epoch 04Epoch 05Epoch 06

  • Data Scientist

    Turns messy data into decisions people trust.

    You design experiments, find the signal in noisy data and explain what it means to the people who decide.

    • Statistics
    • SQL
    • Machine learning
    • Storytelling

    Go deepest on

    Epoch 01Epoch 02

  • Computer Vision Engineer

    Teaches machines to see, from medical scans to robots.

    You build perception systems that detect, segment and understand images and video in the real world.

    • CNNs
    • Vision transformers
    • OpenCV
    • Edge deployment

    Go deepest on

    Epoch 03Epoch 05

  • MLOps Engineer

    Builds the platforms that keep AI reliable at scale.

    You automate training, testing, deployment and monitoring so models improve safely instead of silently breaking.

    • Docker
    • CI/CD
    • Cloud
    • Monitoring

    Go deepest on

    Epoch 06

  • AI Research Engineer

    Pushes the frontier with rigorous experiments.

    You read and reproduce papers, run careful experiments and help turn new ideas into working systems.

    • Mathematics
    • PyTorch
    • Paper reading
    • Experiment design

    Go deepest on

    Epoch 01Epoch 03Epoch 04

How to learn

Learn. Build.
Ship. Share.

The fastest learners treat it like engineering: short loops, real output and public proof. Run this loop for every topic on the roadmap.

A student writing code on a laptop in a calm, minimal workspace
Keep every loop small: one concept, one build, one post.

A sustainable week

Around eight hours alongside your coursework. Consistency beats intensity.

Learn
3 hours
Build
3 hours
Ship
1 hour
Share
1 hour

Total 8 hours a week, every week.

  1. 01

    Learn one idea deeply

    Pick a single concept and stay with it until you can explain it without notes.

  2. 02

    Build it from scratch

    Implement it yourself before reaching for a library. That is where understanding sticks.

  3. 03

    Ship something small

    Put it online, even when it is rough. A live demo beats a folder of notebooks.

  4. 04

    Share what you learned

    Write a short post about it. Teaching is the fastest way to find your gaps.

Questions

Straight answers.

What engineering students usually ask before they start.

Do I need to be great at math to start?

No. You need to be willing to get comfortable with it. Epoch 01 builds the linear algebra, calculus and probability you actually use, with visual explanations first and formulas second. The math tends to click once you see it working inside a real model.

I'm already in my third or final year. Is it too late?

Not at all. The semester labels are a pace, not a deadline. The durations assume six to eight hours a week alongside your coursework. At ten to twelve focused hours a week, you can work through Epochs 01 to 03 in about six to seven months and move straight on to transformers and generative AI.

Can I skip straight to generative AI?

You can build a demo on top of an LLM in a weekend, and you should, because it is motivating. But to debug it, evaluate it and improve it, you need the foundations underneath. Keep a small Gen AI side project going while you work through Epochs 01 to 04 properly.

Do I need an expensive laptop or a GPU?

No. Any laptop that runs a browser and Python is enough for the early epochs. When you need a GPU, Google Colab and Kaggle Notebooks both offer free GPU time, which is plenty for the projects here as long as you keep models small.

Are the resources really free?

Yes. Every course, book, video and guide linked in the roadmap is free to access online. Some platforms sell optional certificates, but you never need one to learn the material.

Should I learn PyTorch or TensorFlow?

Start with PyTorch. It is the default in research and most new open-source work, and the concepts carry over directly if a future team uses a different framework.

How does progress tracking work?

Tick off topics as you learn them and your progress is saved privately in this browser. There is no account and nothing to sign up for. If you switch devices or clear your browser data, your checklist starts fresh.

How do I land an AI internship?

Lead with proof. Two or three well-documented projects, each with a short write-up and a live demo, open more doors than a long list of certificates. Epoch 07 covers portfolios, outreach and interview preparation step by step.

Your AI era starts with one checkbox.

Open the roadmap, tick your first topic and keep going. Future you will be glad you started today.

Start Epoch 01See career tracks
Made with Modulify