Abhinav Agrawal

Building AI products | GenAI expert | Probablistic ML PhD

abhi_profile.png
Currenlty found in Bay Area

Hi there! I recently defended my PhD and am officially in the industry job market. I bring a unique blend of deep technical expertise, zero-to-one product experience, and excellent communication skills. Read more below or set up a short call :)

I recently completed my PhD in computer science at the University of Massachusetts Amherst, advised by Prof. Justin Domke. My work leveraged generative modeling approaches for designing efficient sampling methods.

I remain broadly interested in artificial intelligence. Recently, I led product at SigIQ.ai to build India’s leading AI tutor app, where I managed a team of designers, engineers, and AI experts. Prior to that, I spent a wonderful summer designing generative-model-based causal models for the Causality and Machine Learning team at Microsoft Research Redmond. I spent another summer as an applied scientist intern working on deep-learning-based ranking models for the personalization team at Amazon.

Before my graduate studies, I spent four amazing years at IIT Kanpur acquiring a background in computer vision, deep learning, and generative models. During this time, I was mentored by Prof. Piyush Rai, Prof. Tanay Guha, and Prof. Vinay P. Namboodiri (undergraduate thesis advisor).

Aside from work, I enjoy keeping myself fit, reading books, making people laugh, and playing board games. I cherish deep, philosophical discussions, including but not limited to AI and its implications :) Feel free to contact me, or set up a quick call.

Selected Publications

  1. AISTATS ’25
    Disentangling impact of capacity, objective, batchsize, estimators, and step-size on flow VI
    Abhinav Agrawal, and Justin Domke
    In AISTATS, 2025
  2. Preprint
    Understanding and mitigating difficulties in posterior predictive evaluation
    Abhinav Agrawal, and Justin Domke
    2025
  3. NeurIPS ’21
    Amortized variational inference for simple hierarchical models
    Abhinav Agrawal, and Justin Domke
    In NeurIPS, 2021
  4. NeurIPS ’20
    Advances in black-box VI: Normalizing flows, importance weighting, and optimization
    Abhinav Agrawal , Daniel R Sheldon, and Justin Domke
    In NeurIPS, 2020