← Back to the map
Learning Paths
Six learning paths
Choose by goal, not by course number. Each path gives a core sequence and optional courses.
01
LLM / Agent Engineering
From NLP foundations to agent engineering and training language models from scratch. The focus is evaluation, not just demos.
02
ML Research / PhD
Build supervised learning, unsupervised learning, and learning theory first, then go deep in one direction.
03
CV / Generation / 3D
Recognition → generative models → 3D representation and rendering, forming a complete vision and graphics stack.
04
Robotics / Embodied AI
Kinematics and control → autonomous systems → learning and manipulation → embodied research.
05
MLSys / Inference
From ML system design to inference serving, then accelerators and networking.
06
AI4Science / Health
ML foundations → domain data → medical/biological FMs → frontier lectures.