PhD candidate in Electrical & Computer Engineering at Purdue University, advised by Prof. Joseph Makin. I did BS in Electrical Engineering from UET Lahore, Pakistan and worked for approximately 6 years as Instrumentation Engineer before starting PhD.

My PhD is focused on data-driven methods for computational neuroscience problems. Specifically, I have been working on using deep neural networks trained for ASR as encoding models of neurons in the auditory cortex. I have shown that these DNN-based encoding models can explain the spiking activity of a small group of neurons at fine temporal resolution. Using these encoding models, we can understand the tuning preferences of these neurons by formulating the problem as an inverse problem and using a generative diffusion model as a learned prior.

I have also proposed an efficient algorithm for solving general inverse problems using diffusion models, named Restart Posterior Sampling (RePS). In addition, I am working on the use of generative modeling approaches for synthesizing speech from stereo-EEG recordings. The challenging aspect of this problem is the small amount of labeled data.

I am interested in probabilistic generative modeling, inverse problems, and computational imaging, with applications to computational photography and neuroscience.

📢 Recent Updates

• 🔥 New arXiv paper (2025):
  Solving Diffusion Inverse Problems with Restart Posterior Sampling

• 📰 PLOS Computational Biology (2025):
  Deep Neural Networks Explain Spiking Activity in Auditory Cortex

• 🎓 Teaching Assistant — Generative Modeling (Graduate Course) (Fall 2025)

🔬 Publications

• Solving Diffusion Inverse Problems with Restart Posterior Sampling
  arXiv, 2025
  arXiv

• Deep Neural Networks Explain Spiking Activity in Auditory Cortex
  PLOS Computational Biology, 2025
  Paper

🎤 Conferences

• COSYNE 2023 — Montreal, Canada
  Poster: Understanding Auditory Cortex with Deep Neural Networks

🌱 Outside Research

• Reading: Investing, economics, business
• Sports: Cricket, badminton