sketches

Course syllabus

LLM Inference Systems

Learn how a transformer becomes a fast, reliable serving system—from the first attention layer to distributed prefill and decode.

01

Implementing a Transformer

Build a minimal decoder-only language model from first principles.

02

Inference as a Systems Problem

Define what a fast, efficient inference system must optimize.

03

GPU Architecture for LLM Inference

Develop the hardware intuition needed to reason about inference performance.

04

Efficient Attention

Understand how modern attention kernels reduce memory traffic and latency.

05

KV Cache Systems

Manage the dominant source of dynamic memory in autoregressive serving.

06

Serving a Model with vLLM

Study a production serving engine from request arrival to token output.

07

Batching and Scheduling

Schedule competing requests while meeting latency and throughput goals.

08

Faster Decoding and Model Optimization

Reduce the work required to produce each useful output token.

09

Distributed LLM Inference

Scale models and workloads across multiple accelerators and nodes.

10

Disaggregated Prefill and Decode

Separate the two inference phases so each can scale independently.

11

Advanced Serving Workloads

Extend the serving stack to emerging model architectures and workloads.

12

Operating an Inference Platform

Run inference reliably under real production constraints.

13

Capstone: Meet the SLO

Design and optimize a serving system against a fixed workload and GPU budget.