News

I'm joining UC Irvine as an Assistant Professor of Computer Science. I'm recruiting PhD students who apply in Fall 2026 to start in Fall 2027.

Presented Are Large Reasoning Models Interruptible? at ICML 2026.

At CVPR 2026, our paper AutoGaze was selected as a spotlight. I also co-organized the fifth What's Next in Multimodal Foundation Models workshop (MMFM5) and served on the CV4Edu panel.

Gave an invited talk at the Edge AI and Vision Alliance Innovation Forum: Beyond Words: Teaching AI to Converse With the Physical World.

Gave an invited talk at Hippocratic AI: Hallucination by Design: Building Models That Know What to Say When They're Wrong.

Co-presented the ASRU 2025 tutorial Multimodal Speech Modeling: from Understanding to Generation.

At ICCV 2025, co-organized MMFM4 and moderated the What's Next in Multimodal Foundation Models panel. Also gave an invited talk at the University of Tokyo on Multimodal Pragmatics for Foundation Models.

Co-organized the IEEE CAI workshop on Human Alignment in AI Decision-Making Systems: An Inter-disciplinary Approach towards Trustworthy AI.

Our audio captioning system received the Judges' Award at the DCASE 2024 challenge.

Completed my Ph.D. in EECS at UC Berkeley, advised by John Canny, and began a postdoc in BAIR with Trevor Darrell.

GPU Accelerated t-SNE received the Best Paper Award at HPML 2018. The library, tsne-cuda, is still going.