AI product engineer with TypeScript
Add AI to real apps, front to back, in one language.
At the end: Ship AI features users touch: streaming chat in React, answers from your data, Claude behind a gateway, and a copilot for a real company.
Stage 1: Foundations
One language for the whole app.
- JavaScript, inside out: How JavaScript really runs.
- TypeScript for real code: Types that catch the bug before it ships.
- React: Components, state and effects.
Stage 2: AI in the app
The features users see.
- Building with LLMs (TypeScript): Messages, streaming, tools and zod output, on the raw API.
- The Vercel AI SDK: A streaming AI chat in React with the Vercel AI SDK.
- RAG: search that answers: Answers from your own data, with Postgres and pgvector.
- Claude and the Anthropic SDK (TypeScript): Claude the right way: tools, caching, vision and PDFs.
Stage 3: Ship it
Keep it fast, cheap and safe for real users.
- Build an AI gateway (TypeScript): Budgets, fallbacks and caching between your app and the models.
- Evals that catch regressions in TypeScript: Know a prompt change did not break anything.
- AI security: attack and defend (TypeScript): Defend the app from prompt injection and leaks.
Stage 4: Build it for real
AI features for a real company.
- AI support copilot project (TypeScript): A support copilot that answers from policy.
- AI booking agent project (TypeScript): A booking agent that acts, safely.
Stage 5: Interview ready
The bugs interviewers ask about.
- AI engineering traps (TypeScript): AI bugs from production, found and fixed.
- React traps: React bugs from production.