CV

$${\LARGE \textit{vita brevis, ars longa}}$$

Skills

Programming
Python
CPLEX/Gurobi/FICO-Xpress and APIs
Julia/JuMP
GAMS/AMPL
Java
MATLAB (Simulink)
SQL
R
C/C++
Software
PLEXOS
Tableau
AIMMS
PSS®E
AutoCAD
MS Office Suite
Enterprise Architect
Domain Knowledge
Optimization Theories and Applications
Machine Learning and Deep Learning
U.S. Electricity Market Operations
Transmission/Distribution System Operation
Battery Bidding Strategies
Language
English (Bilingual)
Mandarin (Native)
Japanese (Professional Proficiency)

Work Experience

  • 2024.06 - Present
    Senior Software Engineer - Lead Professional
    GE Vernova, Inc.
    • Work for the Market Clearing Engine Division as the lead optimization architect/developer for the energy management system. Report to Maria Tamayo - Senior Manager, Software Engineering.
    • Details
      • Develop ISO-scale unit commitment and dispatch operations, working for national-level energy market clearing engines used by MISO, PJM, SPP, and ISO-NE covering 30+ states' electricity management in the U.S.
      • Conduct research and development on large-scale and distributed mathematical optimization models with practical implementations.
      • Lead on the production-level optimization software development using Java-CPLEX API with best software development practice.
      • Upgrade and optimize the workflow and automatic Jenkins Cucumber Test pipelines between operation codebase, customers, and database.
      • Communicate with clients and lead a seasoned team with technical know-how.
  • 2022.03 - 2024.06
    Senior Optimization Engineer
    Ascend Analytics, LLC
    • Work for the Development Department as the chief and sole optimization architect/developer for company flagship millions-earning software. Report to Vena Kostroun - SVP of Development and Co-founder.
    • Details
      • Develop large-scale mixed-integer programming models, including single/dual battery storage bidding, renewable-dominant energy market clearing operations, and multi-year generation and transmission expansion planning for PowerSIMM™ Suite and Smartbidder™.
      • Lead on the production-level software refactoring from GAMS to Python-FICO Xpress API to achieve better coding efficiency and accessibility.
      • Develop production-level optimization solver tuning and achieve higher than 50% performance boost for large-scale mixed-integer program series with million-level variables and constraints.
      • Upgrade and optimize the workflow and automatic pytest unit test pipelines between operation code and database, accelerating pre-/post-processing within the solution procedure.
      • Develop new features in optimization models to satisfy clients' needs, involving tens of unit modules such as battery storage, hydro operation, multi-energy networks, etc.
  • 2021.09 - 2022.12
    Postdoctoral Research Associate
    Lawrence Berkeley National Laboratory (LBNL)
    • Work for the LBNL Electricity Market and Policy Department. This research position involves the investigation on future clean energy market with environmental impacts via machine learning and optimization. Report to Dr. Jiang Lin - Staff Scientist.
    • Details
      • Predict and analyze the solar generation potential in existing building rooftops by Deep Learning techniques, including Graph Neural Nets.
      • Implement Python programs for automatic I/O data flow with professional simulation software. Develop GUI and SQL databases for open-source power system optimization software.
      • Model and simulate the national power capacity expansion planning and production cost modeling in PLEXOS through 2020-2035, considering the power reliability and greenhouse gas emission.
      • Draft grant proposals for US DOE and California funding agencies, including the California multi-objective power-eco optimization and 2050 carbon neutrality projects.
      • Mentor interns and visiting Ph.D., host research calls, and manage academic seminars.
  • 2020.05 - 2021.06
    Research and Development Fellow (Co-op)
    Midcontinent Independent System Operator (MISO)
    • Work for the MISO R&D Department. This research-based position involves real-time supply chain optimization in 'MISO''s' daily operations. Report to Dr. Yonghong Chen - R&D Senior Consulting Advisor.
    • Details
      • Perform time-series prediction of power generators' startup/shutdown trajectory by employing Gradient Boosting Tree for the offline task and Long Short Term Memory network for the online task.
      • Design the deterministic MISO real-time look-ahead power supply chain optimization routine and implement it in Python-Gurobi API with calibration via the MISO production engine.
      • Query and parse the time-series data of units' startup and shutdown processes by SQL in the MISO private database, with cleaning and restructuring of original data pairs.
      • Deploy the improved start/shutdown curves in the stochastic supply chain optimization model on several typical operation days. The daily production cost saving on average is around 8% (106 level in dollars).
      • Assist in software documentation, team collaboration, and milestone reports.
  • 2019.05 - 2019.08
    Research Engineer (Summer Intern)
    National Renewable Energy Laboratory (NREL)
    • Work for the NREL Power System Engineering Center. This research position builds integrated optimization models between steady-state scheduling and transient-state dynamics, focusing on assessing solar energy's potential in the energy-reserve market. Report to Dr. Bryan Palmintier - Principal Research Engineer.
    • Details
      • Build the multi-timescale optimization framework for short-term market analyses in Python-Gurobi API with unique steady-state turbine physical models for renewables.
      • Conduct techno-economic assessments of solar panels' potentials with built-in energy storage systems in the energy and reserve co-optimization, considering the multi-timescale coordination.
      • Debug the simulation software and perform sensitivity analyses with different parameters while adding new and reliable features with validations to mimic the practical market environment better.
      • Help design the interface between scheduling operations in Python and transient-state dynamics in PSS®E.
      • Responsible for software documentation, team collaboration, and milestone reports.

Project Participation

  • 2019.05 - 2021.09
    Stochastic Optimal Power Flow for Real-time Management of DERs (SLAC)
    • Funded by DOE ARPA-E ($4.35M)
    • In this project, I focused on developing a new stochastic power flow model and solution techniques. I worked on tailoring the large-scale distributed ADMM algorithm and deployed it in the stochastic optimal power flow model with practical market operations. The multi-thread processing was utilized when solving independent stochastic subproblems.
    • I generated stochastic scenario trees based on historical wind and solar data by applying the K-means clustering.
    • Final Tech Report
  • 2018.09 - 2021.11
    Multi-timescale Integrated Dynamic And Scheduling (MIDAS)
    • Funded by DOE SETO ($3.00M)
    • In this project, I focused on developing a generalized software package on Python/Pyomo for multi-timescale short-term market operations interfaced with the system dynamics simulation. Final code availability is dependent of NREL.
    • Focusing on the scheduling part, I designed and realized flexible and user-friendly functions for system operations, such as retrievals of system sensitivities like LMPs. The I/O interfaces between different optimization modules were also implemented.
    • Final Tech Report
  • 2019.10 - 2020.10
    Hybrid Stochastic Energy Storage Management for Integrated Energy Systems
    • Funded by GEIRI-NA ($500K) via Dr. X. Zhang.
    • I developed a power market simulation platform in MATLAB for the multi-energy network, including electricity, gas, and heat, exemplified by the operations of multi-energy storage systems. Nonconvex power flow, gas flow, and heat flow equations are convexified via the second-order cone.
    • I analyzed energy storages' capability of providing different energy products in the steady-state energy market. Scenario-based stochastic programming for a massive amount of distributed renewable resources was employed.
    • Code
  • 2018.01 - 2019.10
    Multi-Stage Stochastic & Robust Power System Capacity Expansion Planning
    • Funded by DOE NETL ($400K)
    • I designed a multi-stage and multi-timescale capacity expansion planning framework considering plant contingencies and renewable uncertainties. I formulated the problem as a hybrid stochastic and robust MILP program with discrete and continuous uncertainties.
    • I devised a Benders-embedded Column-and-Constraint Generation algorithm to facilitate the solution. Extensive economic assessments were carried out to evaluate long-term renewable paybacks.
    • Final Tech Report, Presentation.

Education

Coursework

SMU
Adaptive Algorithms in Machine Learning (A)
Deep Learning and Neural Networks (A)
Stochastic Optimization (A-)
Distribution System Analysis (A)
Data Analytics and Decision Support (A)
Machine Learning in Power Systems (A)
IIT
Statistical Pattern Recognition (A)
Applied Optimization for Engineering (A)
Objective-oriented Programming and Computer Simulation (A)
Power Market Operation and Economics (A)
Control and Operation of Electric Power Systems (A-)
Power System Planning (A)

Honors and Awards

Academic Interests

  • Stochastic Optimization Theories
    • Traditional stochastic optimization including scenario-based multi-stage stochastic programs, robust optimization, and distributionally robust optimization.
    • Newly blossomed (again) approximate dynamic programming including reinforcement learning, online convex optimization and fast solution algorithms.
    • Application of the state-of-the-art deep learning techniques assisting in searching for high-quality solutions of ultra-scale (stochastic) optimization problems, including using GPU and quantum computing units.
  • Electricity Market Operations and Economics
    • Wholesale electricity market clearing processes including unit commitment and economic dispatch under stringent scenarios of 100% renewable penetration.
    • Real-time energy bidding and trading of flexible energy resources like battery storages and distributed energy resources.
    • Capacity planning and market power exercise from bulk renewable companies.