After the core: specialise · The Intelligence and Data Roles · path 7 of 13
AI / ML Engineer
Builds, trains and runs AI models, and turns them into dependable features inside real products.
🔒 This path opens once you pass the final examination. You can read the path guide and module list now.
Choose this path if you want to train and test models that predict, classify or generate, and check that they are fair and reliable.
Where you see it daily: Chat assistants, "you may also like" suggestions, photo search and spam filters.
Builds on: Level 3 (Programming Skills), Level 5 (Databases and Data) and Level 7 (Artificial Intelligence).
Start here
Ready
Path guide
Role, outcomes, prerequisites, setup, costs, hardware and workload. Open to everyone.
Entry diagnostic
10 questions on the core lessons this path builds on, with revision links. Optional.
Projects and capstone
4 briefs with starter files, acceptance tests and a saved self-check.
Completion record
Your lessons, quizzes, path test and project self-checks in one place.
What you'll focus on
Deep learning frameworks
Build and train neural networks: the layered models behind modern AI.
Tools and ideas: PyTorch, TensorFlow, scikit-learn, Hugging Face
MLOps
Treat models like software: track every experiment and dataset version, put models online, and watch how they behave after launch.
Tools and ideas: MLflow, DVC, Triton, FastAPI
Pick a domain
Go deep in one area: large language models and retrieval (RAG), computer vision, or reinforcement learning.
Tools and ideas: LLMs, RAG, computer vision, reinforcement learning, LangChain, vector databases such as Pinecone
What makes this path different
AI and ML systems learn behaviour from data instead of following rules someone wrote. Unlike most paths, results are probabilities, not certainties, so measuring quality, spotting bias and planning for wrong answers are part of the job.
Modules
Opens after you pass the final examination.
- 1ML1: Thinking in machine learningAI / ML Engineer · Focus area 1: Deep learning frameworks✓
- 2ML2: Preparing data without fooling yourselfAI / ML Engineer · Focus area 1: Deep learning frameworks✓
- 3ML3: Classic models with scikit-learnAI / ML Engineer · Focus area 1: Deep learning frameworks✓
- 4ML4: Evaluation you can trustAI / ML Engineer · Focus area 1: Deep learning frameworks✓
- 5ML5: Neural networks and deep learningAI / ML Engineer · Focus area 1: Deep learning frameworks✓
- 6ML6: Language models and retrieval (RAG)AI / ML Engineer · Focus area 3: Pick a domain✓
- 7ML7: Responsible AI: fairness, explanations, privacy and securityAI / ML Engineer · Focus area 3: Pick a domain✓
- 8ML8: Serving modelsAI / ML Engineer · Focus area 2: MLOps✓
- 9ML9: MLOps: tracking, monitoring and retrainingAI / ML Engineer · Focus area 2: MLOps✓
- 10ML10: Capstone: three AI projects for DakDostAI / ML Engineer · Capstone✓
40 questions from every module, in new situations. Opens once you've passed the final examination and every module quiz in this path.
What you'll build
Projects you can show, because working software proves capability in a way tests alone can't.
- A model trained and tested on a realistic practice dataset, with honest accuracy figures.
- A model served behind an API (FastAPI), with monitoring and a drift check.
- An assistant that finds and quotes answers from practice help pages with simple retrieval (no hosted language model) and refuses when it can't, checked against test questions.