Shenyu Lu is a Researcher in the Machine Learning Department at NEC Laboratories America in Princeton, NJ. He received his PhD in Electrical and Computer Engineering from Purdue University and his MS and BS in Electrical and Electronics Engineering from Huazhong University of Science and Technology. Prior to joining NEC Labs, he gained industry experience as a Machine Learning Engineer Intern at DoorDash. His research focuses on trustworthy and robust machine learning, with particular emphasis on developing methods that make AI systems more reliable across diverse and challenging real-world conditions, including addressing spurious correlations in multimodal and vision-language models, improving robustness in multi-task learning, and building AI that behaves dependably in safety-critical settings.

Recent work includes a paper accepted at ICLR 2025 on mitigating spurious correlations in zero-shot multimodal models, advancing the reliability of vision-language models like CLIP without requiring labeled target data, as well as contributions to CVPR 2025 on identifying and mitigating spurious correlations in multi-task learning. By developing machine learning systems that generalize correctly rather than exploiting statistical shortcuts, his research directly supports NEC’s mission to build intelligent, trustworthy AI technologies capable of performing in complex, real-world environments. His work helps extend the frontier toward AI that is not only accurate but also principled, robust, and safe in practice.