|
Mansi Maheshwari Reinforcement Learning & Lifelong Learning Hello! I'm a Master's student in Computer Science at the University of Massachusetts Amherst where I am fortunate to be advised by Professor Bruno Castro da Silva at the Autonomous Learning Lab. My research focuses on lifelong reinforcement learning, studying how RL agents can continually adapt under non-stationarity. This work has been published at CoLLAs 2025 (poster) and has been accepted at AAMAS 2026 (oral). I also shared this work with a broad audience through my Three Minute Thesis! In Summer 2025 and Spring 2026, as an AI Research Intern at CNH, I built a scalable multi-task vision system for driverless tractors. Grounded in real-world hardware and safety constraints, this experience motivated my Master's thesis, which extends my RL research to embodied settings, co-advised by Professor Hao Zhang. Alongside my research, I am deeply committed to broadening participation in AI. I taught Fundamentals of AI to high school students as an AI Instructor at the University of Washington and am currently consulting with iCEV to help design an upcoming AI textbook for secondary education. Previously, I earned my B.S. in Electrical Engineering from the University of Washington and spent two years as a Software Engineer at Nordstrom. I will graduate in May 2026 and welcome the opportunity to discuss research and explore potential collaborations, especially in lifelong reinforcement learning for socially meaningful applications. |
|
NewsMarch, 2026Guest speaker at Cohere Labs, presenting a talk on addressing the plasticity-stability dilemma in RL. December, 2025Our paper has been accepted for oral presentation at AAMAS 2026. September, 2025Began Master's Thesis co-advised by Prof. Bruno Castro da Silva and Prof. Hao Zhang. September, 2025Consulting with iCEV on an upcoming AI textbook for high school students. August, 2025Presented AltNet at CoLLAs 2025 (Conference on Lifelong Learning Agents). June, 2025AI Instructor at the University of Washington, co-designing and teaching the Fundamentals of Artificial Intelligence course for high-school students. May, 2025AI R&D Intern at CNH Industrial, developing multi-task perception models for driverless tractors. April, 2025Selected finalist for Three Minute Thesis (3MT) at UMass, "From California to Alaska: Teaching AI to Adapt." September, 2024Joined University of Massachusetts Amherst as an M.S. student in Computer Science. ResearchHuman goals, values, and environments evolve over time, yet most AI systems are trained once and deployed as static artifacts. This mismatch limits long-term human–AI collaboration. My research focuses on moving beyond static, task-specific models toward adaptive agents that learn over long horizons. At UMass, I studied lifelong reinforcement learning in simulated robotics tasks, investigating continual adaptation under non-stationarity. At CNH, I conducted research in multi-task supervised learning, exploring shared representations in vision system for driverless tractors. Recognizing the sim-to-real gap and motivated to ground my work in real-world impact, my Master's thesis extends these ideas to physical robots, where long-term adaptation must be both embodied and structured. See my Master's Thesis Proposal for details.
|
|
Instructor, Fundamentals of Artificial Intelligence – University of Washington |
|
Artificial Intelligence Subject Matter Expert – iCEV |
|
Template adapted from Jon Barron's site. |