Learning path
Quantum machine learning and optimisation
“I work in AI or operations research and want to know where quantum really helps.”
Data scientists, ML engineers and optimisation specialists. Assumes gradient descent and a loss function are familiar.
3 milestones · 12 steps · 78 minutes of reading, plus hands-on time
By the end, you will be able to
- Build and train variational circuits, and diagnose barren plateaus
- Formulate an optimisation problem for QAOA
- Benchmark a quantum model honestly against a classical baseline
- Separate genuine quantum advantage claims from dequantisable ones
Milestone 1: Circuits as models
4 stepsYou can read a parameterised circuit as a function with trainable weights.
- Auto-tracks
- LessonIntermediate12 minGates & Circuits
Rotations are your weights; entangling gates are your layers.
Auto-tracks - ChallengeIntermediateTilt the coin
Set a rotation angle to hit a target distribution — training, by hand.
Auto-tracks - LessonBeginner10 minInterference: Where the Power Comes From
The mechanism any quantum model must exploit to beat a classical one.
Auto-tracks
Milestone 2: Variational algorithms
4 stepsYou can train a hybrid quantum-classical loop and know when it will stall.
- LessonAdvanced16 minNISQ Algorithms and VQE
The variational principle, and what near-term hardware can do with it.
Auto-tracks - VisualizationAdvancedVQE and the Barren Plateau
Watch an optimiser descend — and then watch the barren plateau flatten it.
- Auto-tracks
Milestone 3: Honest benchmarking
4 stepsYou can make — and refute — a quantum ML claim with evidence.
- LessonAdvanced16 minQuantum Machine Learning
Kernels, variational classifiers, and where the advantage arguments live.
Auto-tracks - VisualizationAdvancedThe HHL Algorithm
The famous linear-systems speedup, and the fine print that limits it.
Progress is stored in this browser only. Lessons tick themselves when you finish them and challenges when you solve them; everything else you tick yourself, because opening a page is not the same as understanding it.
After this path
- Run a variational loop against real hardware with parameter sweeps and runtime sessions through the API.
- Hold every quantum ML result you read to the classical-baseline standard you practised here.