Malla Siddharth Reddy
I study how AI models store, update, and lose knowledge, and how ideas from the brain can help them learn better.
In the brain, new learning can interfere with old memories, and the brain has ways to manage that. AI models face the same problem: changing one fact can quietly disturb others. I want to understand how these interactions happen inside models, so we can design learning that adds new knowledge without breaking what is already there, and build AI agents that keep improving from experience.
I also work in the other direction: using AI to understand the brain itself.