Yash Mali
Email: ymali@student.ubc.ca | Google Scholar | LinkedIn | GitHub | CV
About
π I am an undergraduate student at the University of British Columbia (UBC) broadly interested in Machine Learning/Artificial Intelligence and Optimization. Specifically, I am interested in interpretability, understanding generalization, and uncovering interesting mechanisms about how small and large models work under the hood. In the past, I have worked on applying AI across a wide range of scientific and social domains.
Awards
Advanced Machine Learning Network: AML-TN
April 2025
βAML-TN sponsored internships highlight the value of developing young researchers as the next generation of machine learning specialists.β
2X Undergraduate Research Award: WLIURA
May 2024, 2025
βThese awards subsidize professors to hire international undergraduate students to work full-time on their research projects in the Summer Session (May to August).β
Interpretability
Undergraduate Research | Jan 2026 β Present
Investigating interpretability in domain adaptation. Advised by Dr. Evan Shelhammer.
Undergrad Thesis πΎ | Sep 2025 β Present
Researching how data diversity and noise affect what autoregressive models learn and the kinds of solutions they find through a controlled synthetic data setting. Advised by Dr. Christos Thrampoulidis.
AI for Science
Quantum ML β Undergraduate Research π§ͺ | UBC Chemistry | Jan 2026 β Present
Comparing how quantum variational circuits and MLPs encode information differently for organic chemistry applications. Advised by Dr. Jolene Reid.
Healthcare AI β Undergraduate Research π©Ί | UBC Medicine (co-op) | May 2025 β Sep 2025 (Continuing part-time)
NLP tools for patients and doctors. We collaborate with UBCβs Cloud Innovation Centre for certain projects. Current projects include agentic NLP pipelines with UIs hosted on AWS for Bipolar Disorder, Depression and Cancer Care. Advised by Dr. John Jose Nunez.
AI and Automation Developer 𧬠| Lux Bio (co-op) | Sep 2024 β May 2025
Applied AI-based drug discovery tools like AlphaFold and ProteinMPNN to optimize sequences and 3D structures of enzymes. Revamped automation systems for bioprocess engineering, orchestrating sensors, pumps, motors, and valves.
Computer Vision & Automation β Undergraduate Research π¬ | UBC Engineering @ Frostad Research Group (co-op) | May 2024 β Sep 2024
Developed particle tracking software using an ensemble of open-source computer vision models along with a UI to correct mistakes. Automated and developed data collection software for new instruments invented by the research group. Advised by Dr. John Frostad.
Instrumentation | UBC Biological Internet of Things 𧬠| May 2025 β Present
Automating brewing/fermentation equipment with IoT-controlled devices and experimenting with ML for brewing!
Papers and Talks
LookSharp: Attention Entropy Minimization for Test-Time Adaptation
Yash Mali, Evan Shelhamer
We minimize the entropy of CLS-to-patch attention in the final layer as a novel TTA loss, encouraging the model to maintain focused attention on shifted data. We demonstrate that attention entropy minimization improves robustness on ImageNet-C. We also show that it is complementary to output entropy minimization and maintains performance on clean data. Preprint here and code coming soon!
π Accepted to ICLR workshop CAO.
A Chatbot for the Management of Bipolar Disorder
Yash Mali, Zejiao Zeng, Kayoung Heo, Grace Zhang, Jincheng Chen, Kamyar Keramatian, Gayatri Saraf, Marco Solmi, Edwin Tam, Sagar V. Parikh, Ayal Schaffer, Serge Beaulieu, Raymond Ng, Lakshmi N. Yatham, John-Jose Nunez
Posters: CAIDA/TrustML ICML Visits 2025 (July 15, 25β), UBC Psychiatry Research Day 2025 (June 5 25β)
Talk: Multidisciplinary Undergraduate Research Conference 2025 | Mar 21, 25β
Preprint here on medRxiv. Under review at The Canadian Journal of Psychiatry!
Praxis UBC: Developing AI education modules in Faculty of Arts Courses Website: here
Jonathan Graves, Laura Nelson, and and prAxIs Contributors (including Yash Mali)
Feb 10, 26β: Accepted at Conference on Teaching and Research in Economic Education workshop π
Quantification of Starch Gelatinization Properties in Glucose and Sucrose Solutions using ParCS
Lily Santos Oβ Keefe, Yash Mali, John Frostad
Talk: Multidisciplinary Undergraduate Research Conference 2025 | Mar 21, 25β
Jan 14, 26β: Accepted at Food Hydrocolloids! π | Preprint here, Paper here
BC Cancer Summit: AI for Cancer Care | November 20, 25β
John-Jose Nunez, Alen Bates, Yash Mali, Olivia Cook
Initial demonstration of a conversational agent that helps patients find cancer support resources and collecting attendee feedback to guide improvements. Talk agenda here | Page 9
Social AI & Leadership
ML Engineer β Undergraduate Research π‘ | UBC SCARP & ECE (part-time) | May 2025 β Present
Using latest developments in NLP and Computer Vision to analyze public records from Vancouverβs housing development approval process. This work bridges AI and social science to address Canadaβs housing crisis. Advised by Dr. Julia Harten and Dr. Christos Thrampoulidis.
Unpacking AI π¨ | UBC Arts (part-time) | May 2025 β Present
Developing modules in existing faculty of arts courses that highlight how AI can be used in their field. For example, sequence modelling in economics or computer vision in archeology. Funded by UBCβs Teaching and Learning Enhancement fund (TELF). Advised by Dr. Laura Nelson and Dr. Jonathan Graves. More info. Tangentially, working on benchmarking LLMs for historical faithfulness.
UBC AI Club π¦Ύ | Jan 2025 β Present
President: Leading initiatives to encourage student understanding and future pathways in AI and ML.
UBC Uncrewed Aircraft Systems βοΈ | Sep 2024 β Present
Leading the ML sub-team to explore and tune open-sourced models for object detection and tracking. This is a small piece of the puzzle on our drones that compete in two university-level autonomous drone competitions every year.