Choose your adventure
Start with Foundations and Hardware, then branch out. Lessons marked Pro are advanced; the first part of every Pro lesson is free to read.
Local LLM Inference
Course overview →Run models on your own hardware
TRACK 1 · BEGINNER
Foundations
Tokens, context windows, prefill vs decode and sampling: the mental model behind every local LLM.
4 lessons · ~45 min · 4 free
TRACK 2 · BEGINNER
Hardware & Sizing
VRAM math, memory bandwidth, Apple Silicon and how to pick a machine that fits the models you want.
4 lessons · ~48 min · 3 free
TRACK 3 · INTERMEDIATE
Models & Quantization
GGUF, K-quants, AWQ, GPTQ, EXL2 and KV-cache quantization, and what each costs in quality.
5 lessons · ~111 min · 2 free
TRACK 4 · INTERMEDIATE
Runtimes
Ollama, llama.cpp, vLLM and MLX: what each is for and how to run them properly.
5 lessons · ~72 min · 3 free
TRACK 5 · ADVANCED
Performance Tuning
GPU offload, batching, flash attention, speculative decoding and prompt caching.
5 lessons · ~114 min · 1 free
TRACK 6 · ADVANCED
Serving & APIs
OpenAI-compatible endpoints, Docker, concurrency, load testing and production hardening.
4 lessons · ~82 min · 1 free
TRACK 7 · INTERMEDIATE
Building Applications
Local RAG, structured output, function calling and agents on top of your own model.
7 lessons · ~108 min · 1 free
TRACK 8 · INTERMEDIATE
Edge & Browser
Small models, WebGPU in the browser, phones and NPUs.
3 lessons · ~37 min · 2 free
TRACK 9 · INTERMEDIATE
Evaluation & Operations
Measure speed and quality honestly, pick the right model, and handle licenses and privacy.
4 lessons · ~51 min · 2 free
Serving AI in Production
Course overview →From one working model to a reliable service
TRACK 1 · ADVANCED
Production Architecture
Turn a working model into a service: SLOs, capacity planning, cost per token, hardware choices and gateways.
4 lessons · ~56 min · 2 free
TRACK 2 · ADVANCED
Kubernetes for GPU Inference
GPUs on Kubernetes, packaging models, production vLLM manifests, scheduling and multi-tenancy, streaming networking, serving platforms with GitOps, and a hands-on cluster lab.
8 lessons · ~187 min · 1 free
TRACK 3 · ADVANCED
Scaling
Queueing maths, autoscaling policies, cold starts, batch workloads and multi-region capacity, with a simulator lab.
7 lessons · ~161 min · 1 free
TRACK 4 · ADVANCED
Logging, Monitoring and Tracing
Structured logs, log pipelines, Prometheus and Grafana for GPU inference, SLO alerting and OpenTelemetry tracing for RAG and agents, with a full observability-stack lab.
7 lessons · ~173 min · 1 free
TRACK 5 · ADVANCED
Operating at Scale
Autoscaling, observability, safe rollouts, reliability patterns and a monitored deployment lab.
5 lessons · ~116 min · 2 free
TRACK 6 · ADVANCED
Delivery, Reliability and FinOps
CI and canaries for models, incident response and postmortems, multi-tenancy and billing, client-side retries and timeouts, and unit-cost FinOps.
5 lessons · ~72 min · 1 free
AI Security and Privacy
Course overview →Build LLM apps that are hard to abuse
TRACK 1 · INTERMEDIATE
Threats to LLM Apps
The LLM threat model, prompt injection, data leakage, insecure output handling and excessive agency.
6 lessons · ~88 min · 2 free
TRACK 2 · ADVANCED
Defense & Privacy
Layered defenses, PII handling, supply-chain safety, compliance basics and a red-team lab.
6 lessons · ~133 min · 2 free
LLM Evaluation and Testing
Course overview →Know whether your AI actually works
TRACK 1 · INTERMEDIATE
Evaluation Fundamentals
Golden datasets, metrics, statistical honesty and LLM-as-judge done properly.
5 lessons · ~74 min · 3 free
TRACK 2 · ADVANCED
Evaluation in Practice
Benchmarks and contamination, testing RAG and agents, regression tests in CI, online evaluation and a harness lab.
7 lessons · ~150 min · 1 free
Inference Engineering
Course overview →Why inference is fast or slow, and how to make it faster
TRACK 1 · ADVANCED
Bottlenecks and Hardware
Prefill versus decode, the roofline model, attention optimisations and GPU architecture for inference.
3 lessons · ~49 min · 1 free
TRACK 2 · ADVANCED
The Software Stack
From CUDA kernels and compilers to the engines: llama.cpp, vLLM, SGLang, TensorRT-LLM and friends compared from the inside.
2 lessons · ~29 min · 0 free
TRACK 3 · ADVANCED
Performance Techniques
Quantisation, speculative decoding, parallelism, disaggregated serving and caching, with the maths for each.
5 lessons · ~74 min · 0 free
TRACK 4 · ADVANCED
Benchmarking and Capstone
Benchmark honestly, avoid common traps, and run the baseline-to-3x capstone.
2 lessons · ~136 min · 0 free
Build an LLM from Scratch
Course overview →Tokenizer to trained transformer, with code you run
TRACK 1 · INTERMEDIATE
Text to Tensors
Write a byte-pair-encoding tokenizer, then turn token ids into batches and embeddings.
2 lessons · ~36 min · 1 free
TRACK 2 · INTERMEDIATE
Attention and the GPT
Self-attention from first principles, the causal mask, multi-head attention and the full GPT block with exact parameter counts.
2 lessons · ~37 min · 0 free
TRACK 3 · INTERMEDIATE
Training and Generation
Autograd, Adam, a runnable tiny GPT, sampling strategies and the KV cache.
3 lessons · ~98 min · 0 free
TRACK 4 · ADVANCED
Modern Upgrades and Scaling
RoPE, RMSNorm, SwiGLU, GQA, LoRA fine-tuning, scaling laws and the PyTorch mapping.
3 lessons · ~53 min · 0 free
Build a Reasoning Model
Course overview →Measure, scale and train reasoning step by step
TRACK 1 · INTERMEDIATE
Reasoning Foundations
What reasoning means, verifiers, and honest evaluation.
2 lessons · ~33 min · 1 free
TRACK 2 · ADVANCED
Inference-time Reasoning
Chain of thought, self-consistency voting, search and verifier loops with a price on every extra token.
2 lessons · ~37 min · 0 free
TRACK 3 · ADVANCED
Training for Reasoning
Policy gradients and GRPO-style training, distillation, tool use and a cost-versus-accuracy lab.
3 lessons · ~132 min · 0 free
Build an Inference Engine
Course overview →A tiny llama.cpp, written by you
TRACK 1 · ADVANCED
Engine Core
The forward pass end to end, loading weights from GGUF with mmap, and KV cache and sampling inside an engine.
3 lessons · ~61 min · 1 free
TRACK 2 · ADVANCED
Making It Fast
Quantised matrix multiplication, CPU speed techniques, continuous batching and a benchmark against llama.cpp.
4 lessons · ~239 min · 0 free
Multimodal Inference
Course overview →Vision, speech, embeddings and images
Fine-tuning and Customisation
Course overview →Adapt a model without wrecking it