Allen George Ajith

I’m an ML Researcher + Engineer and I like taking things apart to see why they work - which turned out to be a decent way to end up doing AI research.

I completed my Master’s in Computer Science at New York University’s Courant Institute of Mathematical Sciences and did my Bachelor’s in Computer Science and Engineering at CUSAT.

I’m broadly interested in AI alignment and AI for social good. My research is centered around mechanistic interpretability and representation learning in LLMs. I currently work with Shauli Ravfogel on two concurrent threads: scaling analysis of linear truth representations across Llama models, and understanding arithmetic operator overloading through pre-training and layer-wise activation analysis.

Previously, I’ve worked on news sentiment analysis for FX trading, federated learning for privacy-preserving AI, link prediction in social networks, and code-mixed language modeling.

Here is my (hopefully updated) CV