Learning paths and topics

Project paths include exercises and original sources. Topic guides recommend a starting point, a build, and a deeper source.

Project paths

AI & Python

Build a search engine for your notes

Turn a folder of ideas into something you can actually find. Learn how meaning becomes a vector, then put it to work.

Some Python · ~150 minutes · 3 milestones

Web development

Make your first corner of the web

Build a small, thoughtful website about something you love. Make it readable, responsive, and usable with a keyboard.

Beginner · ~120 minutes · 3 milestones

Data & reasoning

Find the story in a small dataset

Ask a question, inspect the evidence, and explain what the numbers can—and cannot—tell you.

Some Python · ~135 minutes · 3 milestones

Topics

Understand transformer models

Start with attention and representations, then connect the ideas to a working implementation.

Learn PyTorch

Learn tensors, autodiff, and a training loop well enough to read modern model code.

Learn retrieval-augmented generation

A path from what RAG is to a small pipeline you can inspect and a paper you can cite.

Choose and use a vector database

Learn the tradeoffs behind embeddings, indexing, retrieval, and production operations.

Evaluate a RAG system

Learn to test retrieval quality, answer quality, and failure cases before trusting a RAG application.

Evaluate language model systems

Learn how to judge answers, compare models, and avoid treating a leaderboard as a test of your product.

Fine-tune a language model

Learn when fine-tuning is appropriate, how to evaluate it, and when retrieval or prompting is a better choice.

Learn agentic AI

Learn when a model should call tools or keep state, and when a single prompt is the better design.

Learn LangGraph

Learn graphs, state, and persistence for model workflows that are more than one prompt.

Learn the Model Context Protocol

Learn how tools, resources, and prompts are exposed to a model through MCP.

What Uber's MCP gateway does

A short picture of how Uber's agents call existing APIs through one MCP front door.

Software

Structure a React application

Choose component boundaries, data flow, and state ownership that keep a React codebase understandable.

Software

Add authentication to FastAPI

Learn the essential security decisions behind sessions, tokens, password handling, and protected routes.

Software

Design a reliable cache

Understand when caching helps, how invalidation fails, and how to measure its impact.

Software

Set up GitHub Actions CI

Build a reliable first workflow that tests every change and gives useful feedback when it fails.

Software

Learn Kubernetes fundamentals

Understand pods, deployments, services, and the operational tradeoffs behind container orchestration.

Foundations

Learn linear algebra for machine learning

Build intuition for vectors, matrices, transformations, and the ideas that make modern ML possible.