Interviews Vector
Open Source · MIT
20 phases
503 lessons

AI Engineeringfrom Scratch

Every algorithm built from raw math before a single framework gets imported. Linear algebra to autonomous swarms — Python, TypeScript, Rust, and Julia.

0 / 503 lessons
0% complete

Curriculum Phases

20 phases from setup & tooling to capstone projects. Click any phase to explore its lessons.

0
Ready

Setup & Tooling

Get your environment ready for everything that follows.

12 lessons9 build3 learn

12 of 12 lessons available

1
Ready

Math Foundations

The intuition behind every AI algorithm, through code.

22 lessons16 build6 learn

22 of 22 lessons available

2
Ready

ML Fundamentals

Classical ML — still the backbone of most production AI.

18 lessons16 build2 learn

18 of 18 lessons available

3
Ready

Deep Learning Core

Neural networks from first principles. No frameworks until you build one.

13 lessons13 build

13 of 13 lessons available

4
Ready

Computer Vision

From pixels to understanding — image, video, 3D, VLMs, and world models.

28 lessons27 build1 learn

28 of 28 lessons available

5
Ready

NLP: Foundations to Advanced

Language is the interface to intelligence.

29 lessons24 build5 learn

29 of 29 lessons available

6
Ready

Speech & Audio

Hear, understand, speak.

17 lessons13 build4 learn

17 of 17 lessons available

7
Ready

Transformers Deep Dive

The architecture that changed everything.

16 lessons12 build4 learn

16 of 16 lessons available

8
Ready

Generative AI

Create images, video, audio, 3D, and more.

15 lessons14 build1 learn

15 of 15 lessons available

9
Ready

Reinforcement Learning

The foundation of RLHF and game-playing AI.

12 lessons11 build1 learn

12 of 12 lessons available

10
Ready

LLMs from Scratch

Build, train, and understand large language models.

24 lessons20 build4 learn

24 of 24 lessons available

11
Ready

LLM Engineering

Put LLMs to work in production.

17 lessons16 build1 learn

17 of 17 lessons available

12
Ready

Multimodal AI

See, hear, read, and reason across modalities — from ViT patches to computer-use agents.

25 lessons16 build9 learn

25 of 25 lessons available

13
Ready

Tools & Protocols

The interfaces between AI and the real world.

23 lessons15 build8 learn

23 of 23 lessons available

14
Ready

Agent Engineering

Build agents from first principles — loop, memory, planning, frameworks, benchmarks, production, workbench.

42 lessons36 build6 learn

42 of 42 lessons available

15
Ready

Autonomous Systems

Long-horizon agents, self-improvement, and the 2026 safety stack.

22 lessons22 learn

22 of 22 lessons available

16
Ready

Multi-Agent & Swarms

Coordination, emergence, and collective intelligence.

25 lessons16 build9 learn

25 of 25 lessons available

17
Ready

Infrastructure & Production

Ship AI to the real world.

28 lessons1 build27 learn

28 of 28 lessons available

18
Ready

Ethics, Safety & Alignment

Build AI that helps humanity. Not optional.

30 lessons9 build21 learn

30 of 30 lessons available

19
Ready

Capstone Projects

17 end-to-end products + 9 deep-build tracks. 20-40 hours per project; 4-12 lessons per track.

85 lessons

85 of 85 lessons available

How this curriculum works

Each lesson follows the same loop: read the problem, derive the math, write the code, run the test, keep the artifact. No five-minute videos, no copy-paste deploys. Content is sourced from the open-source ai-engineering-from-scratch project by Rohit Ghumare (MIT license).