Hello, I’m Ishan!
I recently graduated with a BS in Computer Science from Stanford University and am now continuing as a MS student. Currently, I am doing research under Prof. Christopher Ré as part of the Hazy Research Lab.
My research focuses on developing efficient, interpretable methods for large language models, with three core directions:
Benchmarking & Evaluation — Developing rigorous methods to assess foundation model capabilities, including WONDERBREAD, a benchmark for evaluating multimodal models on business process management tasks.
Model Routing — Creating efficient approaches for inference-time model selection, such as Smoothie, a label-free approach for LLM routing.
System Design — Building scalable multi-model architectures that maximize both efficiency and performance in real-world deployments.
I’m excited to explore research opportunities and collaborations in both academic and industry settings. Feel free to get in touch via email to discuss research or collaboration!
Publications and Preprints
