Advanced · 8 hours
AI Developer
Build production applications on top of language models.
0 of 5 lessons · +25 XP per lesson
What you will be able to do
- Design prompts and schemas for structured output
- Implement retrieval over your own data
- Evaluate, monitor and ship model-backed features
Lessons
Lesson 1 · 15 min
Working with model APIs
Streaming, tokens, temperature and cost, explained practically.
- Keep keys server-side, always
- Stream long responses for perceived speed
- Set timeouts and retries on every call
Lesson 2 · 14 min
Structured output
Schemas turn a chat model into a dependable API.
- Define output with JSON schema or tool calls
- Validate before you trust
- Fail loudly, then degrade gracefully
Lesson 3 · 18 min
Retrieval over your data
Chunk, embed, retrieve, then cite.
- Chunk by meaning, not fixed length
- Return sources with every answer
- Measure retrieval quality separately from generation
Lesson 4 · 15 min
Evaluation
Without evals you are guessing at every prompt change.
- Build a small golden dataset early
- Score correctness and format separately
- Run evals in CI before shipping prompt changes
Lesson 5 · 14 min
Production concerns
Rate limits, abuse, privacy and observability.
- Rate limit per user and per key
- Redact personal data before logging
- Track latency, cost and failure rate per feature
Quiz
Score 70% or more to unlock your certificate. +20 XP on your first pass.
Final assessment
Ship a small feature that returns validated JSON from a model, backed by retrieval over ten of your own documents, with a five-case eval set.
Resources
Certificate
Certificates are earned, not given. You need all three:
- ⬜ All 5 lessons completed
- ⬜ Quiz passed (70%+)
- ⬜ Final assessment submitted
Each certificate gets a unique ID, a public verification page and a QR code — so employers can check it in seconds.