Rafael Pastrana
︎ arpastrana@princeton.edu
︎ @arpastrana
︎ @arpastrana
︎ Publications
I am a Research Fellow at the Princeton Laboratory for Artificial Intelligence.
My work focuses on accelerating the mechanical design of meter-scale, architectural structures, such as buildings, towers, and bridges, by combining machine learning, scientific computing, and differentiable simulations. I develop open-source, software tools like JAX FDM and COMPAS CEM.
Most recently, in 2025, I completed my doctoral dissertation at the Form Finding Lab, under the supervision of Forrest Meggers and Sigrid Adriaenssens.
Before that, I worked as a software engineering intern at Robert McNeel & Associates over the summers of 2023 and 2024, developing new technology to solve inverse problems on NURBS curves and SubDs in Rhino.
In 2019, I was at the Block Research Group, and created a digital framework to streamline the fabrication of a 1:1 mockup of the funicular slabs in the NEST HiLo. In 2018, at Gramazio Kohler Research, and in collaboration with Autodesk Research, I trained reinforcement learning models to automate the robotic assembly of timber structures.
Between 2014 and 2017, I worked at Bollinger + Grohmann as a structural engineer. There, I was responsible for the structural analysis and geometric rationalization of several art installations in Europe and in Australia.
I hold a Ph.D. from Princeton and a Masters from ETH Zurich.
︎ @arpastrana
︎ @arpastrana
︎ Publications
I am a Research Fellow at the Princeton Laboratory for Artificial Intelligence.
My work focuses on accelerating the mechanical design of meter-scale, architectural structures, such as buildings, towers, and bridges, by combining machine learning, scientific computing, and differentiable simulations. I develop open-source, software tools like JAX FDM and COMPAS CEM.
Most recently, in 2025, I completed my doctoral dissertation at the Form Finding Lab, under the supervision of Forrest Meggers and Sigrid Adriaenssens.
Before that, I worked as a software engineering intern at Robert McNeel & Associates over the summers of 2023 and 2024, developing new technology to solve inverse problems on NURBS curves and SubDs in Rhino.
In 2019, I was at the Block Research Group, and created a digital framework to streamline the fabrication of a 1:1 mockup of the funicular slabs in the NEST HiLo. In 2018, at Gramazio Kohler Research, and in collaboration with Autodesk Research, I trained reinforcement learning models to automate the robotic assembly of timber structures.
Between 2014 and 2017, I worked at Bollinger + Grohmann as a structural engineer. There, I was responsible for the structural analysis and geometric rationalization of several art installations in Europe and in Australia.
I hold a Ph.D. from Princeton and a Masters from ETH Zurich.