M.S. Student · EECS · MIT (CSAIL + LIDS)
Anakha Ganesh
I study stochastic optimization and learning dynamics, with an emphasis on gradient-free black-box objectives and the inductive bias of modern learning systems.
About
Researching optimization under real-world constraints
I am a Masters student in EECS at MIT (CSAIL and LIDS), advised by Professors Devavrat Shah and Martin Wainwright. My research focuses on stochastic optimization methods for objectives that include gradient-free black-box functions, and on understanding the inductive bias of what learning systems ultimately learn. I am broadly interested in machine learning and statistical inference.
Previously, I completed a double major in Mathematics (18) and Computer Science and Engineering (6-3) at MIT. Please click around to explore my projects on GitHub, past research on Google Scholar, and download my resume below. In my free time, I enjoy pickup soccer, learning about evidence-based wellness, cafe-hopping, and spending time with friends and family.
Currently Working On
Focused investigations in stochastic optimization
Industry Research
Stochastic optimization with Generali
Exploring optimization methods in collaboration with Generali on real-world insurance problems and black-box objectives.
Learning Dynamics
Edge of stability and inductive bias
Studying the edge of stability phenomenon to understand the inductive bias of what is learned during training.
Past Work
Applied research and software engineering
Quantitative Finance
Low-rank covariance modeling
Used low-rank covariance matrices to better characterize economic regimes for portfolio optimization.
Medical AI
Sybil model evaluation
Analyzed Sybil to understand how the CNN characterized lesions as malignant or benign.
Industry
Amazon PXT software development
Built internal tools and workflows on the People Experience and Technology team at Amazon.
Resume