Ashmita Bhattacharyya
I am a fourth-year Engineering Science student at the University of Toronto, majoring in Machine Intelligence, with a certificate in Engineering Business.
Currently, I am working as a Machine Learning Research Intern at Huawei on the LLM Reasoning team, developing transformer-based architectures for ARC-AGI benchmarks.
Prior to this, I was an Undergraduate Researcher at the Vector Institute in CleverHans Lab, working on differentially private ML, supervised by Prof. Nicolas Papernot.
I had also spent a summer at the National University of Singapore, working on an applied ML project on optimizing perovskite solar cell design, in collaboration with the Solar Energy Research Institute of Singapore (SERIS).
I am still early in my career and figuring out where my work will take me, but some consistent areas of interest have been trustworthy AI, general-purpose reasoning, and machine learning for scientific discovery. I am also actively learning agentic systems, both as a tool for automation to make my own work more efficient, and as an emerging research direction.
Growing up, I travelled a lot. I spent most of my childhood in Singapore, a couple years in Dubai, and moved to Montréal in High School. Moving between cities and cultures has shaped me into someone who adapts quickly and connects easily with people from different backgrounds.
Publications
One Step Forward and K Steps Back: Better Reasoning with Denoising Recursion Models