Education
-
Ph.D. in Electrical and Computer Engineering (2017–2025)
University of Illinois Chicago, USA
Thesis: Induced Model Matching: Learning from Restricted Models -
B.S. in Electrical Engineering (2013–2017)
LUMS School of Science and Engineering, Lahore, Pakistan
Courses Taught
- Machine Learning and Deep Learning
- Optimization and Convex Programming
- Scientific Computing and Numerical Methods
- High-Performance and GPU Computing
biography
Usama Muneeb is a researcher in machine learning and theoretical foundations of deep learning. His work lies at the intersection of machine learning theory, natural language processing, reinforcement learning, and scientific computing.
He completed his PhD in Electrical and Computer Engineering at the University of Illinois Chicago, where his research introduced Induced Model Matching, a novel framework that leverages restricted models to improve the training of full-featured models. This work was accepted as a Spotlight presentation at NeurIPS 2024, one of the premier conferences in machine learning and AI.
Alongside his academic research, he has substantial industry experience, including software engineering roles at NVIDIA and Seagate Technology, where he worked on deep learning libraries, GPU-accelerated systems, and resource-efficient neural networks.
Usama is also an active open-source contributor and maintainer, leading key multiprecision semidefinite programming tools used within CVXPY as well as the broader Python ecosystem. His work bridges theory, systems, and practice, with strong engagement in both academic and industrial research communities.
Research Interests
- Machine learning and deep learning theory
- Natural language processing and reinforcement learning
- Scientific computing and numerical optimization
Selected Publications
-
Muneeb, U., & Ohannessian, M. I. (2024). Induced Model Matching: Restricted Models Help Train Full-Featured Models.
Advances in Neural Information Processing Systems (NeurIPS). Spotlight presentation (top 3% of submissions). -
Muneeb, U., Koyuncu, E., Keshtkarjahromi, Y., Seferoglu, H., Erden, M. F., & Cetin, A. E. (2020).
Robust and Computationally Efficient Anomaly Detection using Powers-of-Two Networks.
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
Research & Industry Experience
-
Researcher, University of Illinois Chicago (2017–2025)
Machine learning theory, deep learning frameworks, large-scale experimentation, and research dissemination. -
Software Engineering Intern (Deep Learning Libraries), NVIDIA Corporation (2020)
Contributed to cuDNN v8 development; improved test coverage and performance. -
Research Intern (Data & Image Processing), Seagate Technology (2019)
Model efficiency and data augmentation for resource-constrained environments.
Open-Source Leadership & Service
- Maintainer and contributor, SDPA and CVXPY open-source communities (2022–present)
-
Creator and maintainer of SDPA Multiprecision and SDPA for Python, enabling global access to
multiprecision semidefinite programming through Python - Active documentation author and community contributor in scientific computing ecosystems
Teaching & Expertise Areas
- Machine Learning and Deep Learning
- Optimization and Convex Programming
- Scientific Computing and Numerical Methods
- High-Performance and GPU Computing
