Hi, I am Shahnawaz. I am a deep learning researcher exploring how large neural networks can be compressed to run on edge device at Embedl. I work with research and development of deep learning model optimization techniques, including quantization, pruning, neural architecture search and knowledge distillation. My work at Embedl involves optimizing models so that they can be run on devices such as Qualcomm’s Snapdragon chips, NVIDIA’s Jetson AGX Orin. I also conduct workshops to train deep learning engineers on model optimization strategies, and collaborate with deep learning teams to develop efficient deep learning models.
My Ph.D. was on machine learning for quantum physics applications and exploring the potential of quantum computers for machine learning. I am collaborating with the quantum machine learning team at Xanadu and also working with Dr. Maria Schuld on benchmarking quantum vs classical machine learning. My work involves developing and testing quantum machine learning models on high-performance computing clusters (e.g., NERSC in the US) to assess their properties and capabilities in solving real-world problems. Therefore I have some experience with high performance computing and running large scale machine learning experiments with SLURM and Raytune.
In the past, I worked with Prof. Juan Carrasquilla at VectorAI (now at ETH, Zurich) and Nathan Killoran at Xanadu as a MITACS fellow. I did my master’s thesis in the group of Prof. Franco Nori at Riken, Japan, where I focused on developing numerical approaches to model open quantum systems and explored how deep neural networks can learn the rules of games such as Sudoku.
I completed my Ph.D. at the Wallenberg Centre for Quantum Technology, Chalmers University of Technology, Göteborg. My research focused on the intersection of quantum information, computing, and machine learning. I am interested in exploring the merger of these fields, with a particular emphasis on applying machine-learning techniques to solve problems in quantum physics. In my Ph.D., I developed adversarial neural networks for quantum tomography, a Riemannian optimization technique to speed up quantum process learning, and have written software for quantum physics and circuit simulations during my Ph.D. I have also collaborated with experimentalists to characterize quantum systems, create non-classical quantum states of light (GKP and CAT states) and assist in the data analysis of the first ever measurement of the quantum state of photoelectrons in the group of Nobel prize winning physicist Anne L’Huillier at Lund University.
I am passionate about open-source software and I was actively involved in the team developing QuTiP, the quantum toolbox in Python. Additionally, I have contributed to the PennyLane software for quantum machine learning. I have also been a mentor to students in Google Summer of Code and have myself worked on developing code for the Python software Dipy as a Google Summer of Code 2016 student mentored by Prof. Ariel Rokem at the University of Washington. Please contact me at shahnawaz.ahmed95@gmail.com to discuss collaborations, research opportunities, or any other topic related to quantum physics, machine learning, and software development. You can also connect with me on LinkedIn quantshah. Please visit my Google Scholar page to explore my research publications.
For a detailed overview of my experience and qualifications, you can look at my CV.
Thank you for visiting my GitHub page!