Skip to content

AI & LLM Systems Architecture

Hardware profiling, transformer architectures, and high-performance inference engines.

Curriculum

  • Stanford CS336: Language modeling from scratch, scaling laws, and hardware accounting.
  • Inference Optimizations: Continuous batching, speculative decoding, and KV-cache PagedAttention.
  • Benchmarking Labs: Arithmetic intensity, Roofline analysis, and FP16/BF16 memory benchmarks.