Learning Path
Build understanding one experiment at a time. Follow the full path or choose the skills your role needs.
From a dataset to a dependable system.
Start the learning pathAlready building?
Follow the engineering architecture track from synthetic examples to Qwen baseline review and a local training recipe.
Open Qwen workflowBrowser experiments are educational. GPU training runs in your own environment.
Lessons
30 lessons · 8 hr · 0 marked complete
Data and foundationsPrepare data and build the mathematical and machine-learning foundation.0/7 complete 0/7 complete
01Python as a data toolFoundations · 12 min 02NumPy, Pandas, and data preparationFoundations · 16 min 03Vectors, probability, and gradientsFoundations · 20 min 04Splits, leakage, and model selectionFoundations · 12 min 05Regression, classification, and treesMachine learning · 16 min 06Clustering and self-supervisionMachine learning · 20 min 07Bias, variance, and error analysisMachine learning · 12 min
Model mechanicsTrace neural networks, representations, and transformer behavior.0/6 complete 0/6 complete
08Forward passes and backpropagationDeep learning · 16 min 09Training loops and experiment trackingDeep learning · 20 min 10Autoencoders and latent spacesRepresentation learning · 12 min 11Tokens, IDs, and positionsTransformers · 16 min 12Queries, keys, values, and causal masksTransformers · 20 min 13Architectures and token generationTransformers · 12 min
LLM engineeringBuild retrieval, tool use, and tuned model artifacts.0/5 complete 0/5 complete
14Retrieval, chunking, and grounded answersLLM engineering · 16 min 15Structured outputs, tools, and operational limitsLLM engineering · 20 min 16Prompting, RAG, and supervised tuningFine-tuning · 12 min 17Full tuning, LoRA, QLoRA, and packagingFine-tuning · 16 min 18Validation and catastrophic forgettingFine-tuning · 20 min
Learning from feedbackConnect reinforcement learning, preferences, and alignment.0/5 complete 0/5 complete
19States, actions, rewards, and returnsReinforcement learning · 12 min 20Temporal differences and Q-learningReinforcement learning · 16 min 21Policy gradients, actor-critic, and PPOReinforcement learning · 20 min 22Preference data, reward models, and DPOAlignment · 12 min 23RLHF, RLAIF, principles, and rejection samplingAlignment · 16 min
Agents to productionDesign governed agents, evaluation, and dependable runtime controls.0/7 complete 0/7 complete
24The agent and its harnessAgentic systems · 20 min 25Context engineering and memoryAgentic systems · 12 min 26Data agents and semantic contractsAgentic systems · 16 min 27Tools, identity, and security boundariesAgentic systems · 20 min 28Read a framework through its boundariesAgentic systems · 12 min 29Metrics, graders, and release gatesEvaluation · 16 min 30Serving, monitoring, and rollbackProduction AI · 20 min
Bring this lab to your team Original teaching examples · No production training claims