Hi, I'm Yixin 👋

I work on LLM systems, evaluation, and GPU-accelerated ML infrastructure.
RA @ UCSD Hao AI Lab | M.S. Computer Science student
📍 San Diego, CA | Focused on large-scale training, inference, and agent evaluation.

🏆 GitHub Achievements:

"A journey of a thousand miles begins with a single step." — Confucius

---

💬 Random Dev Joke

---

🔬 Research & Systems Interests

• LLM evaluation & benchmarks (agents, games, scientific reasoning)
• Large-scale training & inference systems (FSDP, vLLM, Ray, Slurm)
• GPU efficiency, memory systems, and model parallelism
• Reinforcement learning for agents (GRPO, NeMo-Gym)

Tech Stack:

---

🛠 Selected Projects

🎮 GamingAgent ⭐ 843 LLM/VLM gaming agents and model evaluation through games → long-horizon reasoning, memory & perception harnesses (Doom, Sokoban, Tetris, Pokémon Red)	🔬 VideoScience ⭐ 5 Benchmark for scientific correctness in text-to-video models → physics & chemistry concepts, VLM-as-Judge scoring (CVPR submission)
🤖 NVIDIA NeMo Gym ⭐ 603 Build RL environments for LLM training → scalable RL training, reward profiling, GRPO Integrating Sokoban & Tetris	🌐 lmenv LLM environment framework for interactive evaluation → standardized interfaces for game-based agent testing

---

---

🧠 Current Focus

• 🔄 Scaling agent evaluation with interactive environments
• ⚡ Training & serving efficiency on multi-GPUs
• 🎯 Reward modeling and RL for LLM agents

📚 Currently Learning

• Advanced distributed training techniques (FSDP, DeepSpeed)
• GPU memory optimization and profiling
• Large-scale RL systems architecture

---

🔗 Connect with Me

• 🌐 Personal Website: yixinhuang48.github.io
• 🔬 Lab Website: hao-ai-lab.github.io/people/
• 📝 Zhihu: 知乎
• 💬 Discussions: Feel free to open an issue or discussion on any of my repositories!

📈 Profile Summary