Part 2 of 5 in the “5 Essential LLM Optimization Techiniques” series. Link to the 5 techiniques roadmap: Link to Sglang code: %20LLM%20Optimization%20Lecture%202%20Parallelisms In this episode, we dive into Mixture of Experts (MoE) and three major forms of parallelism — Tensor Parallelism (TP), Data Parallelism (DP), and Expert Parallelism (EP). Learn how modern LLM architectures like DeepSeek scale efficiently using MoE routing, tensor sharding, and distributed inference strategies.