I'm thrilled to share that my analysis paper about knowledge selection for RAG is accepted by EMNLP 2025 Findings! The result is inspiring for many RAG scenarios, avoiding potential failures. arxiv.org/abs/2410.13258 #emnlp

🚨 The concept of Single Patient Record and integtaion of AI in it In the keynote talk “Health in the AI Era”, Dr Charlotte Refsum proposed the concept of the modern health system design for representing and sharing access to digital health records (DHR). 📸 The figure below

📣 Our University of Utah’s CS Department is hiring about six new faculty members this year! Application links: All areas: lnkd.in/emdaWgPF Data Visualization: lnkd.in/e8zXJtbe Computer Vision: lnkd.in/ev_a9kF4 We welcome everyone to apply! 😀

Google Gemini Code Wiki generates docs from GitHub URL. It can create interactive docs, visualizes code, and lets you chat with Gemini Chat Agent to understand any public repo. 100% free right now.

🚀 I’m on the job market for full-time roles! My PhD research focuses on AI safety & alignment, spanning safety evaluation (LLMs/MLLMs/agent), agent red-teaming, and LLM post-training (SFT & RL). If you have relevant opportunities, I’d love to connect! Recent selected works:

Banger paper for agent builders. Multi-agent systems often underdeliver. The problem isn't how the agents themselves are built. It's how they're organized. They are mostly built with fixed chains, trees, and graphs that can't adapt as tasks evolve. But what if the system could

Claude Code can now run agents asynchronously. Huge for productivity. You can run many subagents in the background to explore your codebase. Work continues uninterrupted. When subagents complete tasks, they wake up/report to the main agent. Workflows feel faster already!