Every topic in AI and ML engineering
87 topics and 1237 hours, grouped by subject rather than by route, because most topics sit on more than one. Every one has a full syllabus behind it.
Programming
- Python33h
The language every AI job is conducted in.
Foundation · AI Engineer, ML Engineer, Research
- Git & collaboration8h
Branches, reviews, and not losing work.
Foundation · AI Engineer, ML Engineer, Research
- HTTP & APIs8h
Requests, status codes, retries, rate limits.
Foundation · AI Engineer
- Async & concurrency10h
Waiting on many slow calls at once.
Core · AI Engineer
- Typing & validation11h
Checking that model output is what you claimed.
Core · AI Engineer
- Data structures & algorithms40h
Still the gate at large companies.
Core · AI Engineer, ML Engineer, Research
- Linux & the shell10h
Where the training runs and the servers live.
Foundation · AI Engineer, ML Engineer, Research
- Testing10h
Pinning behaviour that must not change.
Core · AI Engineer, ML Engineer
- Debugging9h
Narrowing until only one explanation survives.
Core · AI Engineer, ML Engineer, Research
Mathematics
- Linear algebra23h
Matrices, and what a matrix multiply is doing.
Foundation · ML Engineer, Research
- Probability & statistics27h
Distributions, expectation, and reading an experiment.
Foundation · ML Engineer, Research
- Calculus for ML17h
Derivatives, and the chain rule that powers training.
Core · ML Engineer, Research
Data
- SQL & data access14h
Getting the data out, correctly.
Foundation · AI Engineer, ML Engineer
- NumPy & pandas18h
Array and table manipulation, vectorised.
Foundation · ML Engineer, Research
- Data pipelines18h
Moving and reshaping data reliably, on a schedule.
Core · ML Engineer
- Data quality & labelling9h
Where ground truth actually comes from.
Core · ML Engineer, AI Engineer
- Embeddings5h
Text as coordinates, so similarity becomes distance.
Core · AI Engineer, ML Engineer
- Vector databases11h
Approximate nearest neighbour at scale.
Core · AI Engineer
- Retrieval-augmented generation8h
Fetch the relevant text, then answer from it.
Core · AI Engineer
- Chunking & document processing8h
The unglamorous half of retrieval.
Core · AI Engineer
- Reranking & hybrid search10h
Retrieve broadly, then rank precisely.
Advanced · AI Engineer
- Imbalanced and messy data9h
The case where accuracy lies loudest.
Core · ML Engineer
Classical ML
- Supervised learning20h
Learning a mapping from labelled examples.
Foundation · ML Engineer, Research
- Model evaluation & metrics16h
Choosing the metric that matches the cost of being wrong.
Foundation · ML Engineer, Research
- Overfitting & regularisation10h
Memorising the training set instead of learning.
Core · ML Engineer, Research
- Feature engineering11h
Shaping inputs so the model can use them.
Core · ML Engineer
- Unsupervised learning16h
Structure without labels.
Core · ML Engineer
- Time series17h
Data where order carries the meaning.
Advanced · ML Engineer
- Recommendation systems19h
Matching people to items at scale.
Advanced · ML Engineer
Deep Learning
- Neural networks18h
Stacked linear layers with non-linearities between them.
Core · ML Engineer, Research
- Backpropagation17h
How a network learns from its errors.
Core · ML Engineer, Research
- PyTorch30h
The framework research and production both use.
Core · ML Engineer, Research
- Transformers & attention26h
Every token looking at every other token.
Advanced · AI Engineer, ML Engineer, Research
- Pretraining17h
Where the capability actually comes from.
Frontier · ML Engineer, Research
- Fine-tuning18h
When prompting is not enough, and it usually is.
Advanced · AI Engineer, ML Engineer
- Alignment & RLHF22h
Turning a text predictor into something useful and safe.
Frontier · Research, ML Engineer
- Convolutional networks21h
Exploiting the fact that pixels near each other are related.
Core · ML Engineer
- Sequence models before transformers8h
Why attention was invented.
Core · ML Engineer
- Transfer learning11h
Starting from someone else's training run.
Core · ML Engineer
- Hyperparameter tuning10h
The knobs that decide whether training works at all.
Core · ML Engineer
- Reinforcement learning29h
Learning by acting and being scored.
Frontier · Research
- Diffusion & generative models20h
Generating by repeatedly removing noise.
Frontier · Research
Language Models
- Tokenization10h
How text becomes numbers, and why it costs what it costs.
Foundation · AI Engineer, ML Engineer, Research
- The Messages API6h
Stateless request in, one message out.
Foundation · AI Engineer
- Prompting10h
Specification, not incantation.
Foundation · AI Engineer
- Sampling & decoding9h
Temperature, top-p, and why zero is not deterministic.
Foundation · AI Engineer
- Structured output10h
Constrain the shape; the values are still your problem.
Core · AI Engineer
- Streaming responses9h
Perceived latency, not real latency.
Core · AI Engineer
- Context windows8h
Big is not the same as usable.
Core · AI Engineer
- Prompt caching9h
Stop paying twice for the same prefix.
Advanced · AI Engineer
- Choosing a model10h
Matching capability to the task, and re-checking.
Core · AI Engineer
- Multimodal input8h
Documents, screenshots and audio as first-class input.
Advanced · AI Engineer
Agents
- Tool calling7h
The model asks; your code acts.
Core · AI Engineer
- The agent loop14h
A while loop with tool dispatch and a stop condition.
Core · AI Engineer
- MCP & tool interfaces8h
A standard way to expose tools to any model.
Core · AI Engineer
- Multi-agent systems10h
Splitting work across agents, and when not to.
Advanced · AI Engineer
- Context management8h
What to keep when you cannot keep everything.
Advanced · AI Engineer
- Agent reliability8h
Degrading instead of collapsing.
Advanced · AI Engineer
Evaluation
- Evaluation35h
Knowing whether your change helped.
Advanced · AI Engineer, ML Engineer, Research
- LLM-as-judge10h
Grading at scale, once you have proved the grader.
Advanced · AI Engineer
- Eval-driven development10h
Evals in CI, with a train and test split.
Frontier · AI Engineer
- Benchmarks & contamination8h
Why leaderboard numbers mislead.
Frontier · Research, ML Engineer
Systems
- Quantization & inference optimisation19h
Making a model cheaper to run.
Advanced · ML Engineer, AI Engineer, Research
- Distributed training22h
Training something too big for one GPU.
Frontier · ML Engineer, Research
- GPUs & accelerators16h
Why the hardware shapes every decision.
Advanced · ML Engineer, Research
- AI system design17h
The round you will actually get.
Frontier · AI Engineer, ML Engineer
- Distributed systems17h
The classic fundamentals, still asked.
Advanced · AI Engineer, ML Engineer
- Cloud fundamentals12h
Where it all actually runs.
Core · AI Engineer, ML Engineer
- Compression & distillation10h
A smaller model that keeps most of the quality.
Advanced · ML Engineer
Production
- Tracing & observability8h
Seeing what the system actually did.
Advanced · AI Engineer, ML Engineer
- Cost & latency engineering5h
Same quality, an order of magnitude cheaper.
Advanced · AI Engineer
- Prompt injection & guardrails6h
Untrusted text reaching a model that holds permissions.
Advanced · AI Engineer
- Deployment & serving17h
Getting it running somewhere other than your laptop.
Core · AI Engineer, ML Engineer
- MLOps & model lifecycle17h
Keeping a model working after launch.
Advanced · ML Engineer
- Working with coding agents7h
Increasingly its own interview round, with its own rubric.
Core · AI Engineer
- Containers9h
The same environment everywhere.
Core · AI Engineer, ML Engineer
- Product judgement for AI9h
Designing for a component that is sometimes wrong.
Advanced · AI Engineer
- Shipping and measuring10h
Proving the change helped real users, not just the eval set.
Advanced · AI Engineer, ML Engineer
- Interviewing well20h
The skill that converts all the others into an offer.
Core · AI Engineer, ML Engineer, Research
- Experiment tracking7h
Knowing which run produced which model.
Core · ML Engineer
Research Practice
- Reading papers19h
Extracting the idea without reading every line.
Core · Research
- Experiment design18h
Isolating what actually caused the improvement.
Advanced · Research
- Writing & communication16h
Work nobody can follow has not landed.
Advanced · Research
- Research taste18h
Choosing a problem worth solving.
Frontier · Research
- Interpretability26h
Working out what the network is actually doing.
Frontier · Research
- Safety & evaluation research22h
Measuring and reducing what could go badly.
Frontier · Research
- Scaling laws11h
Predicting capability from compute, data and size.
Frontier · Research