About

Hi, I’m Robin. I build production-grade LLM systems and scalable reinforcement learning frameworks, based in Shenzhen.

My work spans LLM post-training, inference optimization, and distributed training. I run a tech blog — Robin’s AI Lab — where I write about LLMs, RLHF/GRPO, vLLM/SGLang, DeepSpeed, and the Ascend ecosystem.

Tech stack: PyTorch, Ray, vLLM, DeepSpeed, Megatron, veRL, HuggingFace, CUDA, Docker, Kubernetes.

Projects

LLM & AI Systems

• LLamaTuner — Efficient fine-tuning for LLaMA, Qwen, Baichuan, GLM, Falcon
• mini-vLLM — Minimal vLLM implementation for learning modern LLM serving
• mini-sglang — Lightweight SGLang-style inference framework
• ScaleTorch — 4D parallelism (TP, PP, SP, DP) in PyTorch
• LLMEval — Modular LLM evaluation framework
• LLMToolkit — PyTorch toolkit for NLP and LLM development
• LLMReasoning — Reasoning techniques for LLMs

Reinforcement Learning

• Deep-RL-Toolkit — Single-agent RL: DQN, PPO, SAC, TD3, and more
• Deep-MARL-Toolkit — Multi-agent RL: MAPPO, MADDPG, QMIX, MAT

Links

• Robin’s AI Lab — Technical blog
• GitHub
• HuggingFace — GaussianTech
• Zhihu

About This Site

Built with AstroPaper and deployed on GitHub Pages.