Building and deploying Physical AI

Led the integration of advanced data science, machine learning, and artificial intelligence methodologies into Altair's flagship physics simulation software, AcuSolve. This strategic initiative successfully broadened the software’s applicability across diverse industry sectors while significantly enhancing user experience and simulation capabilities. Notable achievements include:

• Developed innovative, efficient, and high-performance algorithms for parameter fitting in physics-based battery models. These advancements enabled precise simulation of complex battery phenomena, including health degradation and thermal runaway, leading to published scientific and technical articles.

• Collaborated extensively with industry end-users to integrate newly developed tools and algorithms into their operational workflows, ensuring practical applicability and delivering tangible improvements through cutting-edge technology.

• Project was launched with the goal of using ML-based flow simulators to reduce design time/iterations in automotive aerodynamics.

• Created and Implemented an Encoder-Decoder architecture to predict airfoil flow fields using Convolutional Neural Networks , by inputing just the airfoil image and flow conditions

• Trained the network to obtain a tool which can generalize well to unseen examples and predict test case flow fields with reasonable accuracy

• Used a modified loss function based on gradient loss minimization to improve flow field predictive accuracy.

• Published results as joint first author.