Aditeya Baral

I am a Masters in Computer Science student at New York University’s Courant Institute of Mathematics, Computing and Data Science, where I work at the CILVR lab and the Computation and Psycholinguistics lab, advised by Tal Linzen, Shauli Ravfogel, and Jackson Petty.

I am fascinated by how we acquire and use languages across diverse contexts, specifically in understanding how linguistic information is transformed into abstract internal ideas – and how this process translates to machines. My goal is to uncover the principles underlying how machines learn, understand, and reason with language. My research primarily focuses on language understanding, reasoning, and interpretability through representation learning, with emphasis on studying the training and inference dynamics of language models – how they encode and represent knowledge, apply it to solve tasks, and how we can control and interpret their internal representations.

Previously, I worked as an Applied Research Scientist Intern with the Redis LangCache team, where I built dense embedding-based semantic caching systems for LLMs and open-sourced large-scale datasets and models for text retrieval and re-ranking. Earlier, I was an Applied AI Engineer at Cisco Webex where I fine-tuned on-prem LLMs to integrate secure and cost-effective AI solutions with the Webex AI Assistant.

I have also worked as an Applied Research Scientist Intern at Intel (VSG) Research, where I was advised by Dr. Anbumani Subramanian and Anay Majee to develop attention mechanisms for vision, aimed at autonomous driving in unconstrained, heterogeneous traffic environments like India. During my undergraduate years at PES University, I worked as a Research Assistant at the Center for Cloud Computing & Big Data, where I was advised by Dr. KV Subramaniam and worked on analysing tail latencies for latency-critical applications ranging from language to vision.

You can read about my current and my published research on my publications page!