Mithilesh Vaidya

Hello there! My name is Mithilesh Vaidya.

I am an AI Researcher at Descript, where I work on audio and video generative models for editing and understanding.

I am broadly interested in anything at the intersection of speech and deep learning.

I plan to use this website as a medium for sharing my projects and other random thoughts.

If in a hurry, you can have a look at my CV or Resume.

I blog about my experiences here.

Updates

  • May 2026Submitted PoDAR to NeurIPS 2026.
  • Feb 2026Co-led and shipped the new Audio Regenerate feature in Descript, which allows seamless word-level audio editing via latent inpainting.
  • Jan 2026Video Regenerate is finally out - give it a try in the Descript app!
  • Jun 2024Joined Descript as an AI Researcher!
  • May 2024Graduated with a Master's in Computer Science from Georgia Tech!
  • May 2024Submitted Switching Subspace Model for Neural Population Analysis to NeurIPS 2024.
  • Oct 2023Gave a short presentation on Anthropic's amazing work on Monosemanticity. Slides can be accessed here.
  • Aug 2023Completed a 3-month internship at Qualcomm!
  • Mar 2023Gave a short presentation on In-Context Learning. Slides can be accessed here.
  • Nov 2022Gave a short presentation on Stable Diffusion. Slides can be accessed here.
  • Aug 2022Joined BRAINML@GT and started working with Prof. Anqi Wu on disentanglement of latent variables in VAEs.
  • Aug 2022Joined Georgia Tech for a Master's in CS!
  • Jun 2022Presented Stage 2 of my Dual-Degree Project titled Assessing Comprehensibility of Children's Read Speech. Report available in Projects section.
  • Apr 2022Accepted an offer from Georgia Tech for a Master's in Computer Science!
  • Jan 2022Our work on Deep Learning for Prominence Detection in Children's Read Speech got accepted at ICASSP 2022!
  • Jan 2022Gave a presentation on Wav2vec 2.0 as part of DAP lab at IIT Bombay.
  • Oct 2021Presented Stage 1 of my Dual-Degree Project titled Deep Learning for Prominence Detection in Children's Read Speech. Report available in Projects section.
  • Oct 2021Submitted our work on Deep Learning for Prominence Detection in Children's Read Speech to ICASSP 2022. Preprint available on arXiv.
  • Mar 2021Started my tenure as an Editorial Board Member for Insight, IIT Bombay's student media body.
  • Jun 2020Completed a Hardware Development internship at APT Portfolio. I blogged about my experience!