Optimizing artificial intelligence offers many advantages for mechanical engineers, including faster, more accurate design and simulation, increased efficiency, reduced development costs through process automation, and enhanced predictive maintenance and quality control.
“When people think about mechanical engineering, they think of basic mechanical tools such as hammers, cars, robots, cranes, etc., but mechanical engineering is very broad,” says Faez Ahmed, chairman of marine use Doherty and associate professor of mechanical engineering at MIT. “Mechanical engineering, machine learning, AI, and optimization play a major role.”
In Ahmed course, 2.155/156 (AI and Machine Learning for Engineering Design), students use artificial intelligence and machine learning tools and technologies for mechanical engineering design, focusing on creating new products and tackling engineering design challenges.
Play the video
Cat Trees to Motion Capture: AI and ML in Engineering Design
Video: MIT Department of Mechanical Engineering
“There are many reasons why mechanical engineers think about machine learning and AI basically driving the design process,” says Lyle Regenwetter, teaching assistant for the course and a doctoral candidate for Ahmed’s Design Computation and Digital Engineering Lab (Decoding).
First offered in 2021, the class attracted students from the laboratory departments, including Mechanical and Environmental Engineering, Aeronautics, Astronauts, MIT Sloan Management School of Management, and nuclear and computer science, making it one of the Faculty of Mechanical Engineering (Meche)’s most popular non-core products.
Published to both undergraduate and graduate students, the course focuses on implementing advanced machine learning and optimization strategies in the context of real-world machine design problems. From bicycle frame design to urban grids, students will take part in competitions related to AI in physical systems and tackle optimization challenges in a class environment that is fostered in friendly competition.
Students are given challenge problems and starter codes that have given them a solution but not the best solution. Mesh graduate student Iran Moyer explains: “Our job was (deciding), how can we do better?” Live Leaderboard encourages students to continually improve their methods.
Em Lauber, a graduate student in systems design and management, says the process has given students space to apply what they are learning and explore practical skills in “literally how to code.”
The curriculum incorporates discussions on research papers, and students also pursue practical machine learning exercises tailored to specific engineering issues such as robotics, aircraft, structures, and metamaterials. For the final project, students are collaborating with a team project that employs AI techniques in designing selected complex problems.
“It’s great to see a variety of class projects and high quality,” Ahmed says. “The student projects in this course often led to research publications and won awards,” he cites an example from a recent paper entitled “Gencad-Self Repairing.” This won the 2025 Best Paper Award for Engineering, Information and Knowledge Management for American Mechanical Engineering Systems.
“The biggest advantage of the final project was that it gave all students the opportunity to apply what they learned in their class to areas where they had many interests,” says Maria Smith, a graduate student at Mesh. Her project was chosen to “Markered Motion Captured Data” and considered predicting the ground strength of runners.
Lauber has adopted a framework of “Cat Tree” design with a variety of modules of poles, platforms and lamps to create customized solutions for individual CAT households, while Moyer has created software to design a new type of 3D printer architecture.
“When you look at machine learning in popular culture, it’s very abstract and there’s a sense that something very complex is happening,” says Moyer. “This class has opened the curtains.”
