Nilesh Kumar Jha,

Assistant Professor

Yardi School of Artificial Intelligence

Indian Institute of Technology Delhi, India

email : nkjha@scai.iitd.ac.in

office : 513A, Academic Complex East, IIT Delhi, New Delhi

phone : 011 2654 8413

AI for Signal Processing

Bayesian model design and inference

Generative models for optimization

6G, Wireless Communication

Learning to Optimize

Digital Twins

Google Scholar

ScAI, IITD

Introduction

Assistant Professor at the Yardi School of Artificial Intelligence (ScAI) at the Indian Institute of Technology Delhi (IITD). Previously, I was a chief engineer with the Advanced Modem team at Samsung Research Institute Bangalore and assistant manager the UPSC-IRSS. I received my Ph.D. from the Hong Kong university of Science and Technology (HKUST) and my Bachelor of Technology (B.Tech) from IIT Delhi from the electrical engineering department.

My research interests include AI for signal processing, generative models for optimization, Bayesian model design and inference.

Education

PhD The Hong Kong University of Science and Technology 2019 – 2025

Supervisor: Prof Vincent K.N. Lau, department of electronics and computer engineering

Thesis: Data-Driven Bayesian Signal Processing for Selected Wireless Communication Applications: From Classical to Deep Learning
Huawei-HKUST R&D Collaboration (2021-2025)

B. Tech Indian Institute of Technology Delhi 2014 – 2018

Department of electronics and computer engineering

Professional Experience

Chief Engineer Samsung Research Institute Bangalore, Bangalore 2025-2026

Advanced Protocols for Modem Design for 6G
Projects in link and system level simulators, waveform design, large wireless models
Ascendant Award (2025), 3rd in ITU AI for 6G challenge in Large Wireless Models.

Research Areas

My research interests covers designing classical and deep generative models, inference algorithms and deep architectures for applications in signal processing across various domains in wireless, audio, image, healthcare, sciences, and robotics. Refer to contact page for collaborations and openings.

Classical Generative Models

Model and design probabilistic graphical models for active/online learning. Such tools serve important role in industrial research and development providing interpretable machine learning designs satisfying latency and computational constraints with uncertainty estimates.

Requires familiarity with streaming/stochastic variational inference, probabilistic graphical models, two timescale stochastic optimization, and relevant domain expertise.

Deep Learning Driven Signal Processing and Inference

Model driven deep learning designs covering digital twins, domain driven deep learning, optimization and physics inspired neural network designs.

Requires familiarity with deep learning architectures, optimization driven designs, and optimization.

Deep Generative Modelling and Optimization

Deep generative models for optimization with intractable distributions. Design of deep latent variable models and optimization with deep utility functions.