💎 Full-Stack Fusion
Bridging the boundaries between AI and HPC, integrating GPU clusters, RDMA networks, compilers, and inference engines.
Building a comprehensive knowledge graph bridging AI and HPC, from underlying chips to large model applications.

Based on open source AI & HPC courses, constructing a complete technology stack from silicon-based computing power to silicon-based intelligence.
GPU/NPU Architecture, Heterogeneous Computing, Memory Hierarchy
SuperPod, IB/RoCE, NCCL Collective Communication
K8s, Docker, Slurm Scheduling & Orchestration
AutoDiff, TVM, PyTorch Core Principles
3D Parallelism, ZeRO, Mixed Precision Training
vLLM, TensorRT-LLM, Quantization & Compression