Skip to main content

AI & Machine Learning

The mathematics that makes models comprehensible, the engineering that makes them usable, and the judgement to know when machine learning is the wrong tool.

3 stages15 steps~510 hours~12 months
Track this roadmap free
  1. 1

    Foundations

    Beginner~12 weeks

    Python, mathematics and the classical models worth knowing.

    • Python for data: numpy, pandas, matplotlib

      Skill~35h

    • Linear algebra and calculus for ML

      Skill~40h

    • Probability and statistics

      Skill~30h

    • Supervised learning: regression, trees, ensembles

      Skill~35h

    • Train and evaluate a model on a real dataset

      Project~25h
  2. 2

    Deep learning

    Intermediate~16 weeks

    Neural networks, and the discipline of honest evaluation.

    • Neural network fundamentals and backpropagation

      Skill~40h

    • PyTorch or TensorFlow to a working standard

      Skill~45h

    • Transformers and modern language models

      Skill~40h

    • Evaluation: leakage, baselines, and why your metric lies

      Skill~25h

    • Fine-tune a model and publish the results honestly

      Project~40h
  3. 3

    Production ML

    Advanced~16 weeks

    Serving, monitoring and the operational reality of models that drift.

    • MLOps: pipelines, versioning, reproducibility

      Skill~35h

    • Model serving, latency and cost management

      Skill~30h

    • Drift detection and monitoring in production

      Skill~25h

    • Responsible AI: bias, evaluation, failure modes

      Reading~20h

    • Deploy a model behind an API with monitoring

      Project~45h