Hi, I'm Raghav.

I'm interested in how modern machine learning models can generalize in non-stationary, adversarial settings, especially feedback-driven environments like markets, by learning the right underlying representations.

More concretely, I care about continual learning: models that update on a stream of data, reuse what they've already learned, and avoid catastrophic forgetting so they stay robust as tasks and distributions shift over time.

My background is in statistics, machine learning, and quantitative research/trading.