Choose Your Path

Three paths, each built for a different stage of your career. Pick the one that fits and start building.

FRESHER

AI Engineering Fundamentals

Outcome: junior AI engineer–ready

Document Chat project — Ship a working LLM tool end to end
Structured Output Extractor project — Turn unstructured text into data your code can trust
RAG Pipeline project — Build retrieval-augmented generation you can actually evaluate
Context Window Manager project — Give your assistant memory that scales without blowing the budget
Multi-step LLM Pipeline project — Chain multiple LLM stages into one reliable pipeline
Tool-Calling Agent project — Give the model tools, not just a prompt
Persistent Agent with Memory project — Build an agent that remembers
Streaming Chat API project — Serve your AI tool as a real, production-shaped API
Prompt Evaluation Framework project — Prove your prompts work — and catch when they stop working
Dockerize Your App project — Package your app so it runs anywhere
Deploy to AWS Lambda project — Deploy to real cloud infrastructure, securely
Full AI Assistant (Capstone) project — Capstone — ship one deployed assistant that uses everything above

MID-CAREER

Outcome: AI in your toolkit

Code Review Bot project — Put AI inside a workflow you already run every day
PR Diff Summarizer project — Generate the PR description your team actually wants to read
Semantic Codebase Search project — Search a codebase by meaning, not just text
Code Documentation Generator project — Automate the documentation work nobody wants to do
LLM Test Generator project — Generate tests that are validated, not just plausible
LLM Observability project — Know what your AI calls actually cost — in money and latency
Incident Summariser project — Turn noisy logs into a report someone can act on at 3am
Team Knowledge Extractor project — Make your team's tribal knowledge searchable
Docs Copilot project — Build a documentation assistant that knows when it doesn't know
SQL Agent project — Let people ask questions in English, safely, against a real database
AI CLI Toolkit project — Unify everything you've built into one real tool
AI PR Reviewer (Capstone) project — Capstone — ship AI that gates real pull requests in CI

ML-FOCUSED

Outcome: own ML systems end-to-end

Local to SageMaker project — Take the model you already know how to train onto AWS
Data Exploration project — Know your data before you trust a model trained on it
Feature Engineering project — Build preprocessing that runs the same locally and at scale
Manual Versioning project — Feel the pain that experiment tracking tools exist to solve
Experiment Tracking project — Replace the spreadsheet with a real model registry
Batch Inference project — Score a million rows without standing up a server
Model Serving project — Stand up a live endpoint and know its real latency
A/B Testing project — Let production traffic decide which model wins
Performance Reporting project — Tell a non-technical stakeholder how the model is actually doing
Model Monitoring project — Catch drift before a customer does
Retraining Pipeline project — Make the system fix itself when the data changes
MLOps Capstone project — Capstone — run the full MLOps loop, observed end to end

None of these quite fit? Answer 4–5 questions about your situation and we'll assemble a custom path from our project library and email it to you.