🧠 ML System Thoughts

Recording the learning path, source code analysis, and architectural thinking of Machine Learning Systems.

---

🚧 Current Working On

Current focus and ongoing exploration.

• Deep in vLLM — Exploring Omni Model and scheduler implementation details

---

📚 Reading Notes

1. Foundations

2. Inference Systems

Architecture analysis of modern LLM inference engines.

• vLLM — PagedAttention, Continuous Batching, scheduler source code analysis

3. CUDA Programming

GPU low-level performance optimization and operator implementation.

• Kernel Implementation — Hand-written CUDA basic operators

4. Training Systems

Distributed training frameworks and optimization techniques.

5. Agent Systems

AI Agent architecture and engineering practices.

---

🎯 About This Project

This is my Digital Garden for ML Systems, focusing on bridging the gap between machine learning algorithms and system engineering.

Focus areas:

• Inference Optimization — Low-latency serving, memory management, throughput optimization
• System Architecture — End-to-end understanding from training frameworks to inference engines