Toward self-improving AI systems
From human-engineered AI to AI-designed AI.
What does it take for AI to improve AI? Cheap compute to serve and train models and agents, capable models trained to perform complex real-world tasks, and autonomous agents that discover and learn from better solutions.
Hover a project to trace how it connects; click it for details.
AI Infrastructure
Efficient and scalable AI systems
Autonomous Optimization
Agents autonomously discovering and learning from better solutions
News
Earlier highlights (4 more)Hide earlier highlights
Publications
Projects I led or co-led, grouped by research area.
Autonomous Optimization
Agents autonomously discovering and learning from better solutions
K-Search: LLM Kernel Generation via Co-Evolving Intrinsic World Model
arXiv preprint · 2026
Model Training
Training capable reasoning models and agents
Sky-T1: Train your own O1 preview model within $450
Findings of EMNLP 2025Featured in The New York Times · The Wall Street Journal · The Information
Earlier work
Networking and distributed systems
Accelerating Data Serialization/Deserialization Protocols with In-Network Compute
Workshop on Exascale MPI, ExaMPI@SC 2022
AdaM: An Adaptive Fine-Grained Scheme for Distributed Metadata Management
International Conference on Parallel Processing (ICPP) 2019Talks
Invited talks & visits.
Qingke Talk Aug '24Blog
All posts →No posts yet.




Recursive
NVIDIA
Jiaolab
ByteDance Seed
Databricks