🏠 Who We Are

We are the AI Infra Team at Xiaohongshu (Rednote), dedicated to building cutting-edge, developer-friendly AI Large Model Full-Stack Infrastructure.

Our team has deep expertise across the complete large model lifecycle — from data to training, compression, inference, and evaluation — forming a robust end-to-end technical loop. We have accumulated strong technical advantages in:

We are active contributors to the AI Infra open-source community, participating in the development of various open-source frameworks. We have open-sourced the Relax RL training framework — try it out and let's build together!

🚀 Our Mission

Accelerating the democratization of AI — making powerful models accessible, lightweight, and developer-friendly for everyone.

We are committed to:

• 🧮 Algorithm Democratization — Making state-of-the-art AI model techniques accessible to all developers
• 🪶 Lightweight Deployment — Enabling efficient model deployment with minimal resource overhead
• 🌍 Open Development — Fostering a collaborative, open ecosystem for AI infrastructure
• 💡 Real-World Impact — Helping teams across every business domain achieve production AI success

Our goal is to empower everyone to unlock the full potential of AI — at lower cost and with greater efficiency.

📦 Open Source Projects

Project	Description	Status
🦾 Relax	A flexible and high-performance Reinforcement Learning training framework

🤝 Get Involved

We warmly welcome collaboration from the open-source community! Whether you are interested in:

• 🐛 Reporting issues or bugs
• 💬 Sharing ideas and feedback
• 🔧 Contributing code or documentation
• 🌟 Starring and spreading the word

Feel free to explore our repositories, open issues, or start a discussion. Let's build the future of AI infrastructure together!