← index

AI Mastery Path

Computer Science for AI

The computing half of becoming an AI expert — the partner to Mathematics for AI. From how a computer works at the level of bits, up through programming, data structures, systems, the GPU, and the Python stack that runs modern machine learning. It starts from zero, and every lesson is here because building AI needs it.

Curriculum · 13 areas, 92 lessons

Part I · How computers work

First principles: how a machine made of switches turns electricity into computation.

• 1Bits and binary — everything is a numbersoon
• 2Boolean logic and logic gatessoon
• 3From gates to a CPU — how a computer computessoon
• 4Memory, storage, and the von Neumann modelsoon
• 5Representing integers, text, and floating-pointsoon
• 6How code becomes execution — compilers and interpreterssoon

Part II · Programming foundations (Python)

Learn to instruct a computer — in Python, the language of AI. Variables, control flow, functions, data.

• 7Your first program — values, variables, and typessoon
• 8Operators and expressionssoon
• 9Conditionals — making decisionssoon
• 10Loops and iterationsoon
• 11Functions — naming and reusing logicsoon
• 12Strings and text processingsoon
• 13Collections — lists, tuples, and dictionariessoon
• 14Reading and writing filessoon
• 15Errors, exceptions, and debuggingsoon

Part III · Data structures

How to organize data so a program can use it efficiently — the containers every algorithm runs on.

• 16Arrays and dynamic listssoon
• 17Stacks and queuessoon
• 18Hash tables — O(1) lookupsoon
• 19Setssoon
• 20Linked listssoon
• 21Trees and binary search treessoon
• 22Heaps and priority queuessoon
• 23Graphs — representing connectionssoon

Part IV · Algorithms

Step-by-step recipes for solving problems — and how to choose a good one. (The analysis math lives in Math 101.)

• 24Algorithmic thinking and correctnesssoon
• 25Linear and binary searchsoon
• 26Sorting algorithmssoon
• 27Recursion and divide-and-conquersoon
• 28Dynamic programmingsoon
• 29Greedy algorithmssoon
• 30Graph traversal — BFS and DFSsoon
• 31Shortest paths and beyondsoon

Part V · Abstraction & program design

The ideas that let you build big systems without drowning — objects, functions, modules, and clean design.

• 32Abstraction — the core idea of computer sciencesoon
• 33Object-oriented programming — classes and objectssoon
• 34Encapsulation, inheritance, and polymorphismsoon
• 35Functional programming — pure functions and map/filter/reducesoon
• 36Iterators and generators — lazy evaluationsoon
• 37Modular design and interfacessoon
• 38Design patterns and clean codesoon

Part VI · Software engineering practice

The craft of building real software: version control, testing, reproducibility — and modern AI-assisted development.

• 39Version control with Gitsoon
• 40Collaborating with branches and pull requestssoon
• 41Writing clean, readable codesoon
• 42Testing and test-driven developmentsoon
• 43Debugging strategiessoon
• 44Environments, packages, and reproducibilitysoon
• 45AI-assisted development with Claude Code→

Part VII · How programs run — systems

What happens underneath the code: memory at runtime, processes, and doing many things at once.

• 46The stack and the heap — memory at runtimesoon
• 47Pointers, references, and memory managementsoon
• 48Processes and threadssoon
• 49Concurrency and synchronizationsoon
• 50Parallelism — doing many things at oncesoon
• 51The operating system's jobsoon
• 52Caching and the memory hierarchysoon

Part VIII · Computer architecture for AI

Why deep learning needs the hardware it does — throughput, parallel matrix math, and moving data fast.

• 53CPUs vs. GPUs — latency vs. throughputsoon
• 54Why GPUs power deep learningsoon
• 55The memory hierarchy and data movement — the real bottlenecksoon
• 56Vectorization and SIMDsoon
• 57Floating-point formats for ML — fp32, fp16, bf16soon
• 58Distributed and accelerated computingsoon

Part IX · Working with data

Getting data in, cleaning it, and querying it — the unglamorous majority of real data and ML work.

• 59Files and data formats — CSV, JSON, Parquetsoon
• 60Relational databases and SQLsoon
• 61Querying, filtering, and joiningsoon
• 62Data cleaning and wranglingsoon
• 63Data pipelines and ETLsoon
• 64Batch vs. streaming, and big datasoon
• 65Data storage at scalesoon

Part X · The Python scientific & ML stack

Where the math becomes code: NumPy, pandas, and PyTorch — the tools you actually build models with.

• 66NumPy — arrays and vectorized computationsoon
• 67Broadcasting and tensor operationssoon
• 68pandas — working with tabular datasoon
• 69Visualization — seeing your datasoon
• 70PyTorch and tensors — the deep learning substratesoon
• 71Automatic differentiation in practicesoon
• 72Running on GPUs in codesoon

Part XI · The internet, APIs & deployment

How software talks across the network — and how a trained model becomes a service people can call.

• 73How the internet works — networks and protocolssoon
• 74HTTP, clients, and serverssoon
• 75APIs — how software talks to softwaresoon
• 76Calling LLM APIs — consuming AIsoon
• 77Building and serving a model behind an endpointsoon
• 78Security, authentication, and the cloudsoon

Part XII · Machine learning systems

The engineering around a model — the workflow, the training loop in code, and getting it into production.

• 79The ML workflow — data to deployed modelsoon
• 80The training loop in codesoon
• 81Overfitting, validation, and generalizationsoon
• 82Evaluation metrics that mattersoon
• 83Experiment tracking and reproducibilitysoon
• 84MLOps — versioning, monitoring, pipelinessoon
• 85Inference, latency, and serving at scalesoon

Part XIII · Inside modern AI

The capstone, where Mathematics for AI and this track meet: how neural networks and large language models actually work.

• 86What a neural network is, computationallysoon
• 87Embeddings — turning everything into vectorssoon
• 88Tokenization — how text becomes inputsoon
• 89The transformer and attention, at a systems levelsoon
• 90How an LLM is trained — data, compute, scalesoon
• 91Inference, context windows, and KV cachessoon
• 92The AI systems frontier — where it's goingsoon