Lead energy optimization modeling efforts for electric vehicle demand response with behind-the-meter micro-grid PV+ BESS for over 50 transit agencies. [Optimization] LP, QP, MILP, MIQP, stochastic programming with Gurobi/CPLEX/GLPK/XPRESS/SCIP solver and Pyomo/PuLP/mpi-sppy framework. [RL optimization approach] Deep-Q Network, Markov Decision Process, Convolutional Neural Net**[Physics-based vehicle battery model] 0RC, 2RC[Visualization] Plotly animation, seaborn, matplotlib. ReactJS/NodeJS**** in development

Developed from scratch MILP optimization algorithm with Pyomo/Gurobi for electric bus fleet charge scheduling (10-100 vehicles) to minimize total cost -- cost of chargers and electricity tariff (TOU $/kWh and demand charge $/kW). Employed techniques such as symmetry reduction and warmstart to reduce run time from hours to minutes.

https://geosci.uchicago.edu/~moyer/Inventory/Chen_etal_Energy_Inventory_2020.pdf• Leading an inventory project documenting all long-lived capital assets along the US fossil fuel supply chain, which covers more than 16 sub-industries totaling more than $10 trillion in upfront cost. Industries include upstream extraction, processing, transportation and usage (including 1 million active wells in the oil and gas extraction infrastructure, 2.6 million miles of pipeline system, and nearly 7000 power plants as of 2012) • Evaluate accurately book values, service life, and age structure. Data comes from more than 1,000 government and private industry sources and requires extensive validation. Communicate with more than 40 industry professionals, academia, and government officials to obtain confidential data and industry insights• Utilize Google Earth Engine and R to analyze large database (e.g., classify more than 1 million records of U.S. oil and gas wells based on location and well production characteristics for accurate cost evaluation)

Modified and debugged open-source computational general equilibrium model CIM-EARTH for US economy-wide simulation study of carbon tax and nuclear power phase out

• Model how capital turnover in Hawaii’s electricity transition can be affected by introducing LNG. • Expand SWITCH (written in Python’s Pyomo package) -- an electricity model framework -- to cover over $200 billion worth of assets (measured in upfront cost): developed new modules to introduce additional optimization constraint; delivered first results in publication quality under minimal supervision and within 5 months

• Built from scratch Google Earth Engine script (a cloud-based computational platform for spatial programming) for nation-wide Indonesia, Malaysia, and Papua New Guinea oil palms planation area expansion and forest loss simulation 1990-2010. • Results submitted (as co-author) for publication in Journal of Advances in Modeling Earth Systems • Developed arcGIS/arcPy/qGIS computer scripts to automate processing of 15 types of biophysical vector and raster layers spanning entire South America for soy supply chain research; completed within 5 months

Worked under Professor Mitchel Small in the Department of Engineering and Public Policy. Developed mathematical model to simulate how information acquisition at different time steps reduces uncertainty, and carry out Monte Carlo simulation via MATLAB scripts

The Center of Atmospheric Particle Studies• Improved data analysis by removing systematic bias in air pollution data measurement for Marcellus Shale impact study• Wrote sections of MATLAB-based scripts to for implementation; goal achieved in 2 months with no prior programming experiences