Supercomputing Systems and AI Lab (SSAIL)

University of Illinois Urbana-Champaign

SSAIL develops systems for efficient, effective, and easy-to-use AI. Our research spans large-model training and inference, reinforcement learning systems, mixture-of-experts, long-context models, scientific AI, and high-performance AI systems.

Research Areas

• Large-Model Training — distributed training, memory optimization, resilience, and elasticity
• LLM Inference — efficient and SLO-aware serving
• Reinforcement Learning Systems — scalable infrastructure for agentic and long-horizon RL
• Mixture-of-Experts — scalable training and communication for sparse models
• Long-Context Models — efficient architectures and systems for long-context AI
• Scientific AI — systems for agentic scientific computing

Research

We release models, datasets, systems artifacts, and interactive demos from SSAIL research projects here on Hugging Face.

🌐 Lab Website: SSAIL 💻 GitHub: Supercomputing-System-AI-Lab