Ojasva Nema

Hi, I’m Ojasva.

I am an undergraduate at IIT Roorkee whose research spans Causal Representation Learning, Deep Generative Models, and Mechanistic Interpretability.

Previously, I was a Research Intern at Yale University with Prof. Harsh Parikh (Causal Evidence & Decisions Studio), working on theoretical frameworks to construct Confidence Sets of Representations and causal representation learning.

At IIT Roorkee, I work at the P-square Lab with Prof. Parikshit Pareek, investigating Bilinear MLPs for architectural inductive bias and surgical unlearning, as well as building foundation models for amortized kernel hyperparameter discovery and physics-aware sparse diffusion models for CERN.

I also lead research on LLM internals at the Data Science Group (DSG), investigating how models route semantic domain knowledge (see our preprint and website). Prior to this, I worked as a Research Engineer at Elimentary (YC F26) engineering quantized local LLM APIs and evaluation infrastructure, and developed Image-Alchemy (ICLR DeLTa workshop 2025).

Off the clock, I’m usually reading about finance (I’m always open to book recommendations!) or watching anime. Always happy to chat about circuits, LLMs, or what’s next in AI!

(And seriously, what would you do if you had infinite compute?)