Advisor: Jonathan Cagan, Levent Burak KaraResearch projects:• Multimodal Pretraining Enables Time-dependent Physics Inference from B-Rep• Self-supervised Learning on CAD for few-shot manufacturability estimation• Topology-Agnostic Graph U-Nets for Scalar Field Prediction on Unstructured Meshes• Curve-based Image Neural Style Transfer• Automating Style Analysis and Visualization with Explainable AI

Instructor: Prof. Levent Burak Kara

• Applied geometric pre-training on CAD encoders as surrogates for time-dependent multi-physics solvers.• Successfully predicted temporal reaction force and nodal displacement field from non-parametric boundary representation.

Instructor: Prof. Jessica Zhang• Delivered recitations and held office hours to help students with MATLAB and numerical optimization algorithms.• Created both programming and written assignments and quizzes solutions and coordinated with 4 graders and 5 TAs.

• Joined the most prestigious installation art company in Taiwan as an intern to explore my passion in art during my job-shifting period.• Installed and maintained art installation on site.• Helped design future installation art in Solidworks.

Uni-directional Tap Detector(UTD) Product Development• Decreased the manufacture process time cost of Uni-directional Tap Detector’s by 75% with automation design.• Design arms’ dimension of Delta Robot to travel in a 5x5x2cm cuboid, with a resolution better than 5μm using 2-phase stepper motors.• Decrease the dark current of UTD less than 1nA.• Enable UTD to endure 1000 hours in a damp heat environment.