Inference Lab

Education

  • Postdoctoral Associate, Massachusetts Institute of Technology, 2023-2025
  • Ph.D., The Ohio State University, 2023
  • M.S., The Ohio State University, 2021
  • B.Sc., Bangladesh University of Engineering and Technology, Bangladesh, 2015

Background

Dr. Alam joined the George W. Woodruff School of Mechanical Engineering as an assistant professor in January 2026. His primary research interests include efficient algorithms and methods for AI-native engineering applications. His group, Inference Lab, focuses on foundational AI for engineering tasks, physics-integrated AI, and robot learning for unstructured manufacturing tasks. Before joining Georgia Tech, he was a postdoctoral associate at MIT. Dr. Alam has received several national and international awards including Google research scholar awards in Applied science in 2024, Ford Best Paper award at IDECTC 2025, SEIKM Best paper award at IDETC 2025, Best paper award at MSEC 2020.

Research

Our research lab creates novel artificial intelligence methods for computational design, robotic manufacturing, and AI-native engineering. We deploy these methods in real-world systems for intelligent, real-time decision-making in unstructured environments. Guided by our motto intelligent integration, we aim to unify design, simulation, optimization, and manufacturing into a single scalable platform, what we call universal engineering intelligence, and some call artificial general engineering intelligence. To achieve this vision, we are working at the bleeding edge of AI and engineering with three specific focus areas:

  • Foundation Models for AI-native Engineering: For decades, engineering software has been built around rigid, unintelligent tools; we are reimagining this foundation by developing the AI-native software stack for engineering tasks. Current foundation models, trained on text and images, have transformed language and vision understanding but cannot reason about the physical world or 3D geometry or rigorous engineering tasks. We envision Vision-Language-Shape (VLS) models, a new generation of foundation models that can unify geometric, physical, and semantic reasoning. These models will connect form, function, and fabrication, enabling AI to truly understand the built world rather than simply describe it.
     
  • Physics and knowledge-integrated AI: Our research aims to build AI that can reason like an engineer. This means teaching machines to understand engineering problems grounded in physics and first principles. We develop novel methods to encode engineering knowledge and physics into generative and multimodal AI systems.
     
  • Embodied Manufacturing Intelligence: Our lab believes that embodiment is an essential part of machine intelligence. Humans develop engineering intuition by engaging with the built world by handling, assembling, and shaping 3D objects. Our research brings this ability to machines. While humanoid and general robot learning aim for open-world generality, our focus on embodied manufacturing intelligence leverages the latent structure of engineering tasks and the built world to achieve both depth of understanding and immediate practical relevance.