FlowMLLab: Open AI for Aerospace Education and Research

Artificial intelligence is changing aerospace engineering, but access to the computing resources, validated data, and hands-on experience needed to use it remains unequal. FlowMLLab (https://github.com/Ehsan-Roohi/FlowMLLab) is an open-source education and research platform at UMass Amherst that helps students connect fluid mechanics, aerospace engineering, and modern machine learning.  

Students do more than run ready-made AI models. They generate and analyze computational fluid dynamics (CFD) data, train and test machine-learning models, examine where predictions succeed or fail, and learn how physical laws can make AI more reliable. Current and planned projects span aerodynamics, propulsion, high-speed and rarefied flows, digital twins, scientific machine learning, and aerospace machine vision.

MinuteFund support will directly help students turn these ideas into working projects. Contributions will provide GPU and cloud-computing access; support student-led independent studies, validation studies, and research demonstrations; and help purchase shared equipment such as NVIDIA Jetson computers, cameras, sensors, data-storage devices, and related experimental hardware. These resources will allow students to move from simulation to physical demonstrations by deploying trained models on compact computers and testing machine-vision methods for flow visualization, feature detection, and experimental measurement.

Support will also help students present their work at technical conferences, workshops, and university research events. These experiences give students an opportunity to communicate their results, receive feedback from the scientific community, and build the professional skills needed for careers in aerospace engineering, computational science, and artificial intelligence.

FlowMLLab already includes tested notebooks that combine Python, CFD, neural networks, reduced-order modeling, and operator learning. New funding will allow us to expand these materials into validated datasets, reproducible notebooks, tutorials, videos, and a free community package that educators, students, and researchers can reuse and extend. Because FlowMLLab is open source, each contribution can benefit more than one student, course, or semester. A project developed today can become a classroom module, a research tool, or a publicly available resource for learners who do not have access to costly commercial software.

Your gift will help transform student ideas into validated simulations, working AI demonstrations, conference presentations, and lasting open-source resources. It will give UMass students the physical insight, computational skills, and responsible AI experience needed to design better aerospace systems while sharing the results with the broader community.

Meet the Team!Dominque HabchiNaoya Miyamoto

Aryan Mohta

Francis Padilla