Hello, I'm

Xiangyi Fan

Computational Optimization Scientist

Specializing in large-scale mixed integer programming and optimization under uncertainty — stochastic, robust, and distributionally robust optimization. Currently at ExxonMobil, developing varied types of MIP plus ML + optimization frameworks for energy production.

View Research Get In Touch

About Me

Xiangyi Fan

Optimization Scientist & Researcher

I am a Computational Optimization Scientist at ExxonMobil in Spring, Texas, where I lead the development of integrated machine learning and optimization frameworks to maximize oil production efficiency.

I earned my Ph.D. in Operations Research & Industrial Engineering from the University of Texas at Austin under Professor Grani Hanasusanto, specializing in optimization under uncertainty including stochastic optimization, robust optimization, and distributionally robust optimization.

My research has been published in top-tier journals including INFORMS Journal on Computing and Transportation Research Part B, and I've presented at major international conferences including INFORMS and International Conference on Continuous Optimization.

Ph.D. Operations Research

UT Austin | GPA: 4.00/4

2018 - 2023

B.S. Transportation Engineering

Tongji University | GPA: 91/100

2014 - 2018

Exchange Student

UC Berkeley

2017 - 2018

Professional Experience

Computational Optimization Scientist

ExxonMobil, Spring, TX June 2023 - Present
  • Lead integrated ML prediction + optimization framework using Difference of Convex proxy models and mixed binary nonlinear programming
  • Mentor interns on MINLP scheduling optimization with facility allocation & rig routing
  • Build MIP models for gradewheel scheduling and chemical product value chain optimization
  • Lead research collaborations with PhD groups at CMU and UT Austin

Mathematical Optimization Scientist Intern

ExxonMobil, Spring, TX May 2022 - Aug 2022
  • Formulated large-scale MIP scheduling model with tens of millions of variables and constraints
  • Designed customized heuristics: aggregate/disaggregate, warm start, rolling horizon, neighborhood search
  • Reduced runtime from 8 hours to 30 minutes through model reformulation and heuristic improvements

Capstone Project - Operations Research Team

Sabre Airline Solutions, UT Austin Jan 2020 - May 2020
  • Built end-to-end framework for airline crew non-availability forecasting with model/parameter selection
  • Developed forecasting approaches: time-series, random forest, and LSTM models
  • Constructed two-stage stochastic manpower model to analyze forecast error impact

Research Intern

National University of Singapore, IORA June 2019 - Aug 2019
  • Developed robust optimization model for product upgrading with capacity constraints
  • Utilized network flow to find extreme points with capacitated lot-sizing constraints
  • Validated robust optimization model reformulation in Python + Gurobi

Research Projects

Data-Driven Distributionally Robust Optimization

INFORMS Journal on Computing

Proposed a method for two-stage DRO problems using ambiguity sets constructed from historical data. Developed semidefinite programming solutions and Benders decomposition algorithms.

DRO Decision Rules Copositive Programming

Transportation Network Redundancy Enhancement

Transportation Research Part B

Built a two-stage mixed-binary nonlinear DRO model for optimal retrofit resource allocation to maximize network redundancy under disasters.

Network Resilience Robust Optimization Benders Decomposition

Distributionally Robust Bilevel Optimization

Working Paper

Developed solution scheme for mixed-binary DRO bilevel optimization with interdiction constraints. Reformulated as binary quadratic program with generalized Benders decomposition.

Bilevel Optimization Interdiction SDP

Oil & Gas Pipeline Network Design

Under Review - Computers & Chemical Engineering

Addressing multiphase fluid-dynamics in optimal design of oil and gas pipeline networks using advanced mathematical programming techniques.

Pipeline Design Fluid Dynamics MINLP

Publications

Published

A Decision Rule Approach for Two-Stage Data-Driven Distributionally Robust Optimization Problems with Random Recourse

INFORMS Journal on Computing, 2024

Improving Transportation Network Redundancy under Uncertain Disruptions via Retrofitting Critical Components

Transportation Research Part B: Methodological, 2025

Under Review

Pre-Disruption Resilient Transportation Network Design: A Compact Mathematical Programming and a Tailored Branch-and-Price Algorithm

Submitted to Transportation Science

Addressing Multiphase Fluid-Dynamics in the Optimal Design of Oil and Gas Pipeline Networks

Submitted to Computers and Chemical Engineering

Stochastic Investment Planning for Multi-Period Production Networks with Production Uncertainty: A Two-Level Decomposition Framework

Submitted to IISE Transactions on Operations Engineering

Conference Presentations

2025 EURO Conference, Leeds, UK
2024 INFORMS Annual Meeting, Seattle, WA
2023 Int'l Symposium on Transport Network Resilience, Hong Kong
2022 INFORMS Annual Meeting, Indianapolis, IN
2022 Int'l Conference on Continuous Optimization, Bethlehem, PA
2022 INFORMS Optimization Society Conference, Greenville, SC
2021 INFORMS Annual Meeting, Anaheim, CA

Skills & Expertise

Programming

Python Gurobi CPLEX MOSEK SQL R MATLAB AIMMS SAS

Analytics

Mathematical Optimization Machine Learning Statistical Modeling Data Analytics Vibe Coding

Research Areas

Distributionally Robust Optimization Stochastic Programming Mixed-Integer Programming Network Design Decomposition Algorithms

Soft Skills

Problem Solving Mentoring Collaboration Technical Communication Presentation

Get In Touch

Let's Connect

I'm always interested in discussing optimization problems, research collaborations, or opportunities in operations research and data science.

fxy@utexas.edu linkedin.com/in/xiangyi-fan +1 510-697-2299