Mason Lee Email and Phone Number

reinforcement learning for wildfire suppression, data-driven wildfire severity modeling, vegetation research

Modeling high dimensional coupled PDE systems such as climate models presents restrictive computational cost. My research explores general physics based constraints in the latent space of deep autoencoders to learn solution operators on low dimensional dynamics models. Exploring reduced order modeling applications within Deep Operator Network (DeepONet) and Neural PDEs frameworks.Advisor: George Karniadakis + Aniruddha Bora

Growth + SWE Ops

Exploration of advanced physics-informed reinforcement learning techniques for active flow control of chaotic, turbulent systems.Advisor: Jenna Eppink + Chung Sheng Yao + Louis Edelman

Accurate mapping of glacier rivers is important for hydrology-based climate models but traditional remote sensing techniques are plagued by discontinuous river networks. I proposed and developed a VGG16-UNET based architecture for river detection and a generative conditional variational autoencoder for (CVAE) for river continuity. These methods exceed traditional morphological operators and will allow climate scientists to conduct reliable hydrology tests at fine resolutions. Advisor: Professor Laurence C. Smith + PhD. Sarah Esenther

Development of a distributed early-warning wildfire alert system built on the ODIN data assimilation architecture in scala + akka. Use of LLM agents to parse unstructured wildfire information into geospatial coordinates for integration into wildfire spread models. Automated monte-carlo methods for predictive fire forecasting and situational awareness.Advisor: Peter Mehtliz + Sequoia Andrade

Wildfire Physics Sensitivity Research. HPC + ML computing. Statistical methods for evaluating ELMFIRE dynamical model sensitivity to vegetation, topographic, and parameter variance on the 2021 Creek Fire, California.Advisor: Nancy Kiang + Robert Field + Keren Mezumann + Marcus van Lier-Walqui

Worked free-lance for various startups assisting in integrating AI into their data pipelines. I focused mostly on systems optimization using deep learning, predictive analytics, and machine learning enhanced remote sensing techniques. Aided several companies in activating data in prepraration of funding rounds.

Leaving over 400 homes destroyed, the CZU Fire devasted the small mountain community of Boulder Creek located near Santa Cruz, California. I worked with mycologists and biologists around the country to develop a biofilter capable of degrading toxic heavy metals before they entered the adjacent waterways. Using native Oyster mushrooms, the CoRenewal team has developed a holistic approach to bioremediation that is non-invasive and in-line with native remediation techniques. Aided experimental design and spearheaded meta-data anaylsis for ongoing research.

Worked on a Type I CALFIRE Handcrew to suppress and control wildfires across California in the worst fire season in the state's history. Responsible for conducting a variety of fire suppression methods such as controlled burns, fireline construction, and water pump operation. Worked 24-hour shifts on fires while ensuring the highest level of safety for fellow crewmembers.

Aided environmental engineers in water quality research and water sampling for L.A. water remediation project.