Scholar
Zain Sarwar
Google Scholar ID: QOsVyPMAAAAJ
University Of Chicago
Citations & Impact
Citations
130
H-index
3
i10-index
3
Publications
8
Co-authors
0
Publications
Resume (English only)
Academic Achievements
- StructMoE: Structured Mixture of Experts Using Low Rank Experts, NeurIPS EMNSLP 2024
- Dense Backpropagation Improves Training for Sparse Mixture-of-Experts, NeurIPS 2025
- Continual Pre-training of MoEs: How robust is your router?, TMLR 2025
- MYCROFT: Towards Effective and Efficient External Data Augmentation, NeurIPS MlforSYS 2024
- Deepfake Text Detection: Limitations and Opportunities, IEEE S&P (Oakland) 2023
- Can Virtual Reality Protect Users from Keystroke Inference Attacks?, USENIX 2024
- Towards a General Video-based Keystroke Inference Attack, USENIX 2023
Research Experience
- Conducting research at the University of Chicago, working on various projects such as designing sparse augmentations to improve model performance and developing structured reasoning and context pruning techniques.
Education
- PhD student at the University of Chicago, advised by Professor Michael Maire.
Background
- A 5th year PhD student at the University of Chicago in the Department of Computer Science. Research interests include advancing the capabilities and efficiency of LLMs, designing sparse augmentations to introduce dynamic, token-level specialization within standard transformers, and developing techniques for structured reasoning and context pruning.
Miscellany
- Contact: zsarwar@uchicago.edu; Social Media: LinkedIn, Github, Stackoverflow
Co-authors