Research
Work at the intersection of privacy, robustness, training dynamics, and large-scale learning systems.
ReSuMe: Retriever-Summarizer Mutual Enhancement via Reinforcement Learning
The Web Conference (WWW) 2026
The Hidden Cost of Modeling P(x): Membership Inference Attacks in Generative Text Classifiers
NeurIPS 2025 Workshop · TMLR Extended Version (Under Review)
SALSA: A Secure, Adaptive and Label-Agnostic Scalable Algorithm for Machine Unlearning
UAI 2025
Enhanced Identification of Pheochromocytoma and Paragangliomas' Genetic Clusters from CT
SPIE Medical Imaging 2025
Detecting Dataset Outliers via Training Dynamics Clustering
UAI 2026 (Under Review)
Multi-Headed Deep Feature Learning for Genetic Clusters Identification from CT
Biomedical Signal Processing & Control (Under Review)
Experience
A progression through scale, research depth, and production responsibility.
Writing
Explorations in representation learning, philosophy, and the mechanics of modern neural networks.
The Memory Erase Button: Identity, Responsibility, and the Value of Suffering
Not a question of pain, but of authorship, what we choose to erase may be the very thing that wrote us.
Read →The Quiet Tension Between Generations: On Advice, Independence, and the Paths We Choose
The tension between generations is not simply about disagreement; it is the quiet negotiation between preserving the meaning of the past and allowing the future to unfold differently.
Read →The Limits of Kindness: Understanding Goodness Through Dostoevsky's The Idiot
A philosophical exploration of compassion without boundaries, examining Prince Myshkin as a study of whether pure kindness can survive within flawed social structures.
Read →Things I've Built
Systems and tools designed to explore ideas deeply and make them usable.
SALSA Unlearning Library
Unified framework for scalable machine unlearning with multiple baselines.
View on GitHub →SIREN - Explainable AI Framework
Interpretability framework for detecting AI-generated images.
View on GitHub →LoRA for Vision Transformers
Lightweight adaptation framework for Vision Transformers using LoRA
View on GitHub →Beyond Research
Resilience, and long-form thinking; Shaped outside the lab.
Strength & Athletics
Mountains & Water
