Guanglei Wang
My name is Guanglei Wang and I am currently a research scientist at Alibaba Group. My research interests lie in mathematical programming-based optimization methods, robust optimization, combinatorial optimization, and their various applications in supply chain management, telecommunications and power systems. Recently, I have been working on developing open-source solvers and modeling platforms with some excellent colleagues. You can find a brief summary of my research experience here. I am also an applied scientist in constrained optimization, heuristics and deep learning.
My main area of expertise is in developing Mixed-Integer Programming (MIP) solvers, with a particular focus on preprocessing and cut generation techniques for MILP, MIQCP, and MISOCP. I have extensive experience in developing MIQP, MIQCP and MISOCP solvers that have consistently outperformed open-source solvers (without relying on problem-specific fine-tuning tricks), and have even surpassed some commercial ones (COPT, CPLEX) by a significant margin. If you would like further details on my work, please don't hesitate to contact me.
In addition to my work on MIP solvers, I have also worked on developing (Deep) Reinforcement Learning-based methods to enhance the efficiency of autonomous outbound calling systems, as well as creating heuristics to solve inventory and supply chain optimization problems such as transshipment and truck hauling. I am the author of AUTO-RO, a robust optimization software that is widely used in the Alibaba economy. Furthermore, I have collaborated on the development of Gravity, an award-winning mathematical programming language, alongside Dr. Hassan Hijazi. Prior to my work on optimization softwares, I worked at Orange Labs Research (Paris), where I tackled virtual network optimization problems.
I am highly enthusiastic about the emerging LLMs (e.g., ChatGPT, GPT4) and their potential to accelerate the development of traditional MIP solver development. This could lead to the creation of a new generation of computing paradigms.
I am actively exploring technical positions in data science (ideally combining optimization, coding and deep learning) worldwide.
Career
Research Scientist and Staff Research Scientist in Damo Academy, Alibaba Group, Hangzhou, China (Nov 2020-now)
Applied Scientist in Digital Supply Chain Team, Alibaba Group, Hangzhou, China (Sep 2018-Oct 2020)
Research Fellow in the Australian National University, Canberra, Australia (Feb 2017-June 2018)
Research Engineer in Orange Labs Research (France Telecom), Paris (June 2013- June 2016)
PhD in Computer Science from Telecom SudParis and Paris 6 University (Nov 2016)
Intern in Orange Labs Research (France Telecom), Paris (July 2012- Dec 2012)
MSc degree in Operations research in the University of Edinburgh (Dec 2012)
B.Eng in Management Science and Engineering in Zhengzhou University (Aeronautics) (June 2011)
Exploiting sparsity for the min-k partition problem, Mathematical Programming Computation, joint work with Hassan Hijazi (2019).
A Lagrange decomposition based branch and bound algorithm for the optimal mapping of cloud virtual machines, European journal of operations research, joint work with Walid Ben-Ameur and Adam Ouorou (2019).
Gravity: A Mathematical Modeling Language for Optimization and Machine Learning, NIPS 2018 Workshop MLOSS, joint work with Hassan Hijazi and Carleton Coffrin (2018).
Mathematical programming methods for microgrid design and operations, Computational Optimization and Applications, joint work with Hassan Hijazi (2018).
Exploiting sparsity for the min-k partition problem, Mathematical Programming Computation, joint work with Hassan Hijazi (2017).
Multipolar robust optimization, joint work with Walid Ben-Ameur and Adam Ouorou. European journal on computational optimization (2016).
Convex and concave envelopes: revisited and new perspectives, joint work with Walid Ben-Ameur and Adam Ouorou, Operations Research letters (2016).
Optimal mapping of cloud virtual machines, joint work with Walid Ben-Ameur and Adam Ouorou, Electronic Notes in Discrete Math (2015).
On-going work
A convergent hierarchy for solving disjoint bilinear problems.
Reduction methods in MIQP and MIQCP: a computational study.
Selected Presentations
Mathematical programming, the Alibaba Group, Hangzhou, China(July 2018)
Multipolar robust optimization, NICTA, Canberra, Australia (March 2017).
Multipolar robust optimization, Brno, APMOD 2016 (8-10 June, Brno, APMOD 2016).
New perspectives in bilinear optimization, Institute Henri Poincaré, Paris (Nov 2015).
Optimal mapping of cloud virtual machines, HuaWei, Boulogne-Billancourt, France (July 2015).
Awards
2021 COIN-OR Cup with Dr. Hassan Hijazi and other fabulous colleagues.
Tutorials
Introduction to mathematical optimisation (The Australian National University, Nov. 2017)
Profession service
Reviewer of journal articles: Mathematical Programming, Mathematical Programming Computation, European Journal of Operational Research, European Journal on Computational Optimization, Operations Research Letters, Computational optimization and applications, Computers & Operations Research.
I like camping with my family. A number of interesting camping areas can be found here.