My research focuses on building efficient audio-centric models and evaluation methods for evolving real-world applications and workflows. I've worked on:

• Efficient audio-language models (Mellow)
• Evaluation metrics that align with human judgments (AURA Score, MACE)
• Benchmarks for audio generation and understanding (FoleyBench, MMAU-Pro)

Currently, I am a founding engineer at a YC-backed startup working on voice agent evals. Previously, I was a masters student at CMU advised by Professors Chris Donahue and Bhiksha Raj, where I worked on models and evaluation methods for audio-based systems. I have also done research internships at MIT and EPFL. Before that, I did my undergrad in electrical engineering at IIT Delhi.