Learn by shipping, not collecting tutorials.
A personalized path from software foundations to production AI, with time estimates, prerequisites, practice tasks and proof-of-skill projects.
- Progress
- 0%
- Topics
- 0/23
- Remaining
- 175h
- At your pace
- 25 wk
Personalize your plan
AI application developer: Build useful LLM products, RAG systems and AI features. The full roadmap contains about 175 focused learning hours, excluding projects.
Your AI engineering path
Select a node to open its lesson plan. Follow the connections from foundations to production.
Best next action
Python for product development
Recommended for ai application developer. Spend about 8 hours, then complete the practice task and save the proof.
- Completed
- Recommended
- To learn
23 topics shown
Stage 01
Software foundations
Write, test and connect the code every AI product depends on.
0/4 complete0%
Stage 02
Machine learning essentials
Learn enough ML to evaluate models and make sound product decisions.
0/4 complete0%
Stage 03
LLM application development
Move from impressive demos to controlled, testable AI features.
0/5 complete0%
Stage 04
RAG, tools and agents
Connect models to knowledge and actions without losing control.
0/5 complete0%
Stage 05
Production AI engineering
Ship systems that remain secure, observable and affordable after launch.
0/5 complete0%
Scroll vertically to move through every stage. Drag the horizontal scrollbar on smaller screens.
Stage 010/4 complete
Software foundations
Write, test and connect the code every AI product depends on.
Stage 020/4 complete
Machine learning essentials
Learn enough ML to evaluate models and make sound product decisions.
Stage 030/5 complete
LLM application development
Move from impressive demos to controlled, testable AI features.
Stage 040/5 complete
RAG, tools and agents
Connect models to knowledge and actions without losing control.
Stage 050/5 complete
Production AI engineering
Ship systems that remain secure, observable and affordable after launch.
Python for product development
Topic plan · Software foundations
Use functions, modules, types, exceptions, virtual environments and packages confidently.
You should be able to
- Write small programs without copying a tutorial
- Structure code across modules
- Handle invalid input and runtime errors
- Practice task
- Build a CLI that accepts a topic and returns a formatted study plan.
- Proof of completion
- A tested repository another developer can run.
Skill coverage
- Software0%
- ML & data0%
- LLM systems0%
- Agents0%
- Production0%
Five checkpoints that prove your ability
Do not wait until the end. Ship one project after each stage and improve earlier work as your skills grow.
After stage 01
AI-ready API client
Build a command-line program that calls a public API, validates input, handles errors and saves structured results.
- Readable Python package
- Environment-based configuration
- Tests for success and failure paths
- Clear README with setup steps
After stage 02
Support ticket classifier
Train a model that routes support tickets into categories and explain where it fails.
- Documented dataset split
- Baseline and improved model
- Precision/recall comparison
- Failure analysis with examples
After stage 03
Structured AI assistant
Build a streaming assistant that produces validated structured output and handles unsafe or invalid requests.
- Prompt contract and examples
- Structured output validation
- Streaming interface
- Small regression evaluation set
After stage 04
Cited research assistant
Build an assistant that searches private documents, cites evidence and uses one external tool safely.
- Document ingestion pipeline
- Retrieval evaluation set
- Answers with source citations
- Tool permissions and failure handling
After stage 05
Production AI capstone
Turn one previous project into a deployed product with users, monitoring, evaluation and cost controls.
- Deployed application and runbook
- Quality and latency dashboard
- Security threat model
- Post-launch evaluation report
Test knowledge, then build something real.
Use quizzes to find weak concepts and the practical AI course for deeper guided exercises.