Kamal Choudhary is at the forefront of efforts to expand the intersection of electrical and computer engineering and materials science by fusing artificial intelligence (AI) with the physical sciences. As an assistant professor in the Department of Materials Science and Engineering with a joint appointment in the Department of Electrical and Computer Engineering, his work bridges quantum mechanics, classical simulations, and machine learning to accelerate the discovery of next-generation materials.
In this Q&A, Choudhary shares what excites him most about the challenges ahead, how his lab is pushing the limits of AI for materials design, and his vision for empowering the next generation of engineers and scientists.
What are the most important or exciting challenges in electrical and computer engineering today?
One of the most exciting frontiers in electrical and computer engineering is the fusion of AI with physical sciences, especially in areas like atomistic materials design and quantum technologies. Building physics-grounded, interpretable AI systems that can reason, predict, and generate solutions for complex real-world problems is both a grand challenge and an incredible opportunity. These intersections offer a path to breakthroughs in energy storage, quantum computing, semiconductors, and biomedical technologies.
Which of these are you currently tackling, and what drew you to them?
At AtomGPTLab, I focus on atomistic materials design using a blend of quantum mechanics, classical simulations, and machine learning. My group develops AI infrastructure to accelerate the discovery of materials for electronics, energy, and quantum applications. What drew me to this challenge is the belief that AI can fundamentally transform the way we explore materials space, from decades-long trial-and-error processes to intelligent, inverse design workflows that are fast, scalable, and accessible.
Is your research focused more on foundational science, practical applications, or both?
My work is grounded in both. On the foundational side, I develop graph neural networks, large language models, and quantum algorithms that incorporate the laws of physics. On the applied side, we integrate these models into platforms such as JARVIS, AtomGPT, and the ChatGPT Material Explorer used by thousands of researchers globally to design materials for real-world devices.
What’s been your most interesting or surprising finding so far?
A particularly exciting result has been the ability of generative AI models trained on chemical formulas/spectra/images alone to propose atomic structures that are realistic and often match known phases. This highlights the potential of AI-driven inverse design, where users can specify target properties and receive synthesizable materials candidates, significantly accelerating the path from idea to experiment.
What are your goals here at Johns Hopkins?
While ChatGPT and similar tools have revolutionized the field of AI, their applicability is limited for physical‑science applications. Materials science, however, lacks a dedicated, accessible GPT‑based assistant tailored to its unique challenges. One project I’m developing is atomgpt.org, an open‑access generative‑AI framework for discovering and designing functional materials using classical, quantum, machine learning and experimental techniques. This tool is designed to help scientists apply AI while keeping important physics principles in mind.
I want to build a strong network of researchers who combine AI, physics, and engineering to solve major challenges in energy, computing, and sustainability. I’m also looking forward to bringing these tools into classrooms and open‑science projects so that students and scientists from all backgrounds can access them worldwide. I’m excited to contribute to the innovative culture of the ECE department at Hopkins!