Work Areas:

• Conquer modeling problems that have not yet been solved, by Luminary or anyone else, from defining the problem out of a complex set of fuzzy constraints to evaluating state-of-the-art machine-learning methods,tobuilding new methods and pushing the technological frontiers of AI for computational science and engineering…
• Devise the models that process and manipulate physical 3D geometries, for engineering design of a limitless range of real objects used by all of us in everyday life; supercharge simulations without compromising reliability and accuracy; and understand the outputs and goals of Luminary’s platform itself…
• Modernize computer-aided engineering by designing, developing, and testing these large-scale models for Luminary’s platform, for engineering design of a limitless range of real objects used by all of us in everyday life, each with its own physical demands and dynamics…

Technologies and Tools:

• In-depth knowledge of AI workloads and architectural tradeoffs, plus testing and validation methods and metrics…
• Strong programming skills as they relate to ML/DL, including debugging, optimizing, and the associated frameworks…
• Deep mathematical skill and intuition, especially as it relates to machine-learning and deep-learning architectures…

We’re looking for someone who has:

• Conducted extensive research in CS, applied mathematics, science, or engineering on the development of ML and DL methods for scientific computation, engineering design, or related topics… 
• Designed models from scratch, optimized them through fine-tuning, and characterized the various tradeoffs relevant to the application context (e.g., performance versus accuracy)...
• Worked with physically meaningful data, preferably from physical-science-based simulations, and performed uncertainty quantification…
• Owned research projects end-to-end, from defining the problem to achieving results…A passion for the convergence of the physical and simulated world, exploiting PINNs/PINOs to disrupt the optimization of things…