Level 2 · M.Sc
MSc Applied AI Systems
Artificial Intelligence · Live · v0
Applied AI systems end to end — mathematical foundations, model architectures, training, and production deployment.
- Deadline
- Next cohort starts 2026-09-07 (AY 2026/27)
- Length
- 42 weeks · 42 phases
- Cortex Credits (CC)
-
126 CC
What are Cortex Credits? - Syllabus
-
View the phase syllabus
- Phase 00 — Foundations
- Phase 01 — Hardware & Computing Substrate
- Phase 02 — Numerical Representation
- Phase 03 — Linear Algebra from First Principles
- Phase 04 — Calculus & Optimization for AI
- Phase 05 — Probability & Information Theory
- Phase 06 — Python for AI Engineering
- Phase 07 — Scalar Autograd from Scratch (`minigrad.scalar`)
- Phase 08 — Tensor Autograd from Scratch
- Phase 09 — MLP, Modules, and Optimizers
- Phase 10 — Initialization, Normalization, Residuals
- Phase 11 — Tokenization Theory + BPE Implementation
- Phase 12 — The Corpus: Designing the Microscopic Dataset
- Phase 13 — Embeddings & Representation Spaces
- Phase 14 — Pre-Transformer Sequence Models
- Phase 15 — Attention from Scratch
- Phase 16 — Positional Encodings
- Phase 17 — Tiny Transformer Block & Mini-GPT
- Phase 18 — Training Loop, Checkpointing, Mixed-Precision Preview
- Phase 19 — Training Dynamics & Debugging
- Phase 20 — Evaluation Harness
- Phase 21 — Inference Internals & Sampling
- Phase 22 — KV Cache: From Math to Memory
- Phase 23 — GPU Architecture Fundamentals
- Phase 24 — CUDA & Triton Hands-On
- Phase 25 — PyTorch Internals
- Phase 26 — Quantization Deep Dive
- Phase 27 — Modern Attention Optimizations
- Phase 28 — Fine-Tuning, LoRA, QLoRA
- Phase 29 — Retrieval-Augmented Generation (RAG)
- Phase 30 — Structured Generation & Constrained Decoding
- Phase 31 — Tool Use & the Model Context Protocol (MCP)
- Phase 32 — Agents: Planning, Memory, Sandboxing (Capstone Application)
- Phase 33 — Inference Serving: From FastAPI to Continuous Batching
- Phase 34 — Observability, Cost & Capacity
- Phase 35 — Distributed Training & Inference
- Phase 36 — Frontier Architectures
- Phase 37 — Security & Safety of AI Systems
- Phase 38 — MLOps
- Phase 39 — Capstone: The Miniature Production System
- Phase 40 — Hardening, Postmortem, "What's Next"
- Phase 41 — Learner Portal: Delivering the Curriculum to Many
- Enrolment prerequisites
-
- A verified account and admissions-committee approval.
Professor: Borja Tarraso — Founder & Chief
Sign in to request enrolment