Phase 1: Foundations
Week 1 — ML Fundamentals & Intuition
- Module 1.1: The ML Mindset – problem‑first framing, data exploration.
- Module 1.2: Linear & Logistic Regression – Titanic challenge, AUC‑ROC, error trade‑offs.
- Module 1.3: ML Landscape & Taxonomy – supervised vs unsupervised, algorithm personalities.
- Module 1.4: Optimization & Learning Theory – gradient descent game, No Free Lunch intro.
Week 2 — Practical ML & Initial Project
- Module 2.1: Feature Engineering & Data Prep – competition & "feature detective".
- Module 2.2: Model Evaluation & Validation – cross‑validation, model court simulation.
- Module A: Kolmogorov Corner – algorithmic information theory & compression challenge.
- Module 2.3: ML Ethics & Real Stakes – demilitarization, "hard choices" simulation.
- Capstone: Build‑a‑Baseline — churn, sentiment transfer, anomaly detection.
Phase 2: Deep Learning & Transformers
Weeks 1‑2 — Neural Network Fundamentals
- NumPy NN from scratch, backpropagation, optimization olympics.
Weeks 3‑4 — Specialized Architectures
- CNNs, RNNs, modern training techniques with visualizers.
Weeks 5‑6 — RL & Efficient Fine‑tuning
- RL foundations, LoRA & adapter‑based tuning.
Weeks 7‑8 — Transformers & Attention
- Self‑attention, transformer build‑up, pre‑trained model adaptation.
Weeks 9‑10 — Capstone & Advanced Topics
- Self‑supervised, generative, multi‑modal learning & architecture defense.
Phase 3: Decentralized Training & AutoML
Weeks 1‑2 — Distributed Training Foundations
- Data/model/pipeline parallelism, gradient traffic control.
Weeks 3‑4 — Communication‑Efficient Training
- DCT & sparsification, EATreeSGD synchronization.
Weeks 5‑6 — Parallel Evolution & Auto‑Tuning
- Seed chase, evolutionary strategies at scale.
Weeks 7‑8 — Decentralized Ecosystem
- Federated learning, fault tolerance drills.
Weeks 9‑10 — AutoML & Self‑Improving Systems
- NAS, continuous updating AI colony project.