Mehak Dhaliwal
Hi there! I’m a Computer Science PhD student at the University of California, Santa Barbara, advised by Prof. Yao Qin.
My research centers on trustworthy, reliable, and accessible AI systems, with a focus on real-world healthcare applications. My current projects include improving the confidence calibration of large language models, analysing their multilingual performance and cross-lingual generalization in post-training, and developing data- and AI- driven methods to improve automated insulin delivery systems for individuals with Type 1 diabetes.
Prior to this, I completed an MS in Computer Science at UC San Diego, working with Prof. Julian McAuley on data-efficient recommendation systems, and a B.Tech. in Computer Science & Engineering at IIT Delhi, where I worked with Prof. Aaditeshwar Seth on fairness in content recommendation. I also spent time as a software engineer in Samsung R&D’s On-Device AI team.
news
| Apr 2026 | New preprint on ArXiv: Exploring the Role of Multilinguality in LLM Post-Training. Takeaway: English is not all you need. |
|---|---|
| Nov 2025 | Our work on nutrition estimation using multimodal LLMs is now published in Nature Communications Medicine. |
| Oct 2025 | Our work on hypoglycemia risk during real-world physical activity in adults with type 1 diabetes is accepted to Diabetes Technology and Therapeutics. |
| Jun 2025 | Started a summer internship with the Alexa International team at Amazon in Bellevue, working on multilinguality in LLMs. |
| Apr 2025 | Presenting NutriBench at ICLR. See you in Singapore! |