Lead Data Scientist|ML Engineer - Stantec
Prior to 2019:
Postdoctoral Associate - Duke University, Civil and Environmental Engineering
Visiting Postdoctoral Associate - Princeton University, Civil and Environmental Engineering
National Institute of Food and Agriculture (NIFA) Fellow - USDA
My research innovates approaches for quantifying patterns of uncertainty for optimal decisions across a variety of topic areas ranging from transportation to flooding and beyond. My interdisciplinary work is at the nexus of machine learning, data science, and the physical sciences. Recently, I created an approach for distilling engineering physics into machine learning features where the features are dimensionless values based on Buckingham Pi Theorem. Ongoing work includes the development of a method for quantifying extremes under time varying, non-stationary conditions, and in turn applying the method to climate change attribution. In transportation, I am investigating how to predict travel demand (on a road by road basis) by representing travel behavior with hydrodynamic analogues that are connected to mean field theory (from statistical mechanics) and machine learning algorithms.
While at Duke and Princeton universities, my research investigated the interconnections between hydrological processes and ecosystem productivity. Hydrological processes of interest included groundwater and rainfall-runoff dynamics, as well as stochastic representations of hydro-climatic variables such as rainfall and streamflow. Work included a novel, exact solution to the Boussinesq equation describing groundwater dynamics, as well as stochastic descriptions of the rainfall-runoff process in both space and time. We were linking such hydrological descriptions to the productivity of ecosystems as determined by the response of the soil-plant-atmosphere-continuum to various environmental factors such as nutrient availability, climate change (e.g., temperature, precipitation), and water scarcity.
In particular, I previously focused on the optimal leveraging of cultivations with different photosynthetic systems (C3, C4, and CAM) so as to maximize productivity while reducing costs related to land degradation, water consumption, and fertilizer application. Towards this end, we developed a framework called Photo3 that consistently compares the carbon and water fluxes for all three plant photosynthetic systems. An improved theoretical understanding of the plant system, including plant interactions with watershed hydrology, will lead to more informed planning decisions for the sustainability of managed ecosystems under changing climate and land use conditions.
As a professional engineer, I'm excited about using stochastic processes, statistical mechanics and machine learning methods to innovate engineering practices.
Stantec
475 5th Avenue, 12th Floor
New York, NY 10017
Phone: +1 212 233-8250
Mark.Bartlett@gmail.com
7 Arbor Crossing
East Lyme, CT 06333
Phone: +1 508 942-4271
Mark.Bartlett@gmail.com
