Vassilis Digalakis Jr

Assistant Professor, Boston University

Operations & Technology Management, Questrom School of Business (primary)

Systems Engineering, College of Engineering (affiliate)

Center for Information & Systems Engineering (affiliate)

Hariri Institute for Computing, Computational Science, and Engineering (affiliate)

PhD in Operations Research, Massachusetts Institute of Technology

Research Interests:

▪ Trustworthy Artificial Intelligence (AI)

▪ Machine Learning, Optimization, Operations Research

▪ Healthcare & Sustainability

Address: Office 617, 595 Commonwealth Avenue, Boston, MA 02215

Contact: vvd [at] bu [dot] edu || vvdigalakis [at] gmail [dot] com

I am an Assistant Professor of Operations & Technology Management (OTM) at the Questrom School of Business at Boston University (BU). I am also affiliated with the Systems Engineering Division (SE) at the College of Engineering, the Center for Information and Systems Engineering (CISE), and the Hariri Institute for Computing, Computational Science and Engineering. Prior to joining BU, I spent time at HEC Paris as an Assistant Professor of Operations Management and the Alexa Spoken Language Understanding team at Amazon Science as a Research Scientist Intern. I completed my Ph.D. in Operations Research at the Operations Research Center at the Massachusetts Institute of Technology (MIT), advised by Prof. Dimitris Bertsimas, and my Diploma in Electrical and Computer Engineering at the Technical University of Crete (TUC), Greece.

Find my CV here.

Research. I work on trustworthy AI: how to design AI systems that behave reliably, transparently, and appropriately for their context of use. I combine machine learning, optimization, and operations research to build such systems and to study how their deployment and governance shape adoption, downstream decisions, and real-world outcomes, particularly in healthcare and sustainability. My research has been published in journals such as Operations Research, Manufacturing & Service Operations Management, and IEEE Transactions on Knowledge and Data Engineering, and has received awards, including the INFORMS Pierskalla Award and the Harold Kuhn Award.

To find out more about my research, click/expand here.

[What is the problem and why does it matters?] AI models are increasingly embedded in decision-making systems, yet most are designed and evaluated in isolation from how they will be deployed, governed, and used in practice. Especially in high-stakes domains such as healthcare and sustainability, context—including operational and regulatory constraints—shapes trust, adoption, and impact. This gap limits how much value organizations can reliably extract from AI in practice.

[What is trustworthy AI?] Trustworthy AI refers to AI systems that meet core principles such as accountability, transparency, fairness, safety and robustness, privacy, human oversight, and societal and environmental well-being. These principles define what it means for AI systems to be deployable and usable in practice, and they are increasingly formalized through regulatory frameworks such as the EU AI Act and the US AI Bill of Rights.

[What is my approach?] I view trustworthiness as both a modeling and a systems problem. From a modeling viewpoint, I build models that are inherently trustworthy by design—with properties such as interpretability or stability chosen to match the needs of the application. From a systems viewpoint, I study how deployment and governance decisions shape downstream behavior and outcomes. Methodologically, I use tools from optimization, operations research, and machine learning to develop theory and algorithms, and I study their performance once embedded in real operational processes, including (but not limited to) healthcare delivery and sustainability planning.

Figure: List of 20 most common keywords from my first 10 papers

Below are selected research directions that illustrate how I study trustworthy AI, from model design to deployment and real-world use. To find out more, click/expand on each of them.

Trustworthy AI Model Design: Building Structurally Stable Machine Learning Models

[What is the problem and why does it matters?] Many machine learning models change sharply when data, features, or retraining conditions change, even when predictions improve only marginally. In high-stakes settings, this instability undermines interpretability, trust, and usability, limiting adoption in practice.

[What is my approach?] We study structural stability as a modeling objective: designing models and retraining procedures that evolve smoothly over time while preserving key analytical structure. This work spans sparse regression, decision trees, retraining sequences of models, and time-series models, using optimization to explicitly trade off predictive performance and stability.

Slowly Varying Regression under Sparsity (with D. Bertsimas, M. Li, O. Skali Lami). Operations Research, 2024.
Improving Stability in Decision Tree Models (with D. Bertsimas).
Towards Stable ML Model Retraining via Slowly Varying Sequences (with D. Bertsimas, Y. Ma, P. Paschalidis).
Interpretable Time Series Autoregression for Periodicity Quantification (with X. Chen, L. Ding, D. Zhuang, J. Zhao).

Trustworthy AI Deployment and Governance

[What is the problem and why does it matters?] Models that perform well on benchmarks often fail to deliver value once deployed, due to costs, constraints, and compliance requirements. Organizations need principled ways to select and manage models under real deployment conditions.

[What is my approach?] We develop frameworks that treat model selection and deployment as optimization problems over performance, cost, and compliance constraints. This work connects algorithmic design with governance and operational decision-making, clarifying how deployment choices shape adoption and downstream outcomes.

ML Compass: Navigating Capability, Cost, and Compliance Trade-offs in AI Model Deployment (with Ramayya Krishnan, Gonzalo Martin Fernandez, Agni Orfanoudaki).
The Challenger: When Do New Data Sources Justify Switching Machine Learning Models? (with Christophe Pérignon, Sébastien Saurin, Flore Sentenac).

Trustworthy AI for Healthcare and Sustainability

[What is the problem and why does it matters?] Healthcare and sustainability applications require AI systems that are not only accurate, but also reliable, interpretable, and robust to uncertainty. Failures in these domains carry significant human, social, and environmental costs.

[What is my approach?] We design optimization- and learning-based models that embed trustworthiness directly into decision support systems, and study their behavior once deployed in real operational pipelines. Applications include healthcare delivery and public-sector decision-making, as well as sustainability and energy planning under uncertainty.

Decarbonizing OCP (with D. Bertsimas, R. Cory-Wright). M&SOM, 2024.
Selected as the 2023 M&SOM Practice-Based Research Competition Finalist
- Where to locate COVID-19 mass vaccination facilities? (with D. Bertsimas, A. Jacquillat, M. Li, A. Previero). Naval Research Logistics, 2021.
  Won the 2024 Harold Kuhn Award

For a full list of publications and preprints, please visit my Google Scholar page.

FOR STUDENTS: I am looking for motivated students to work together on a variety of projects related to trustworthy AI methodologies and responsible AI deployment. If you are interested, please expand and read below.

Prospective PhD students

If you are interested in pursuing a PhD under my supervision, please apply to the BU Questrom PhD Program, Operations & Technology Management track, and mention my name in your application: https://www.bu.edu/questrom/phd-program/

Current BU PhD students

If you are already a PhD student at BU and interested in collaborating, please send me an email.

If you are interested in working with me as a Research Assistant (RA):

Please send me an email with the subject line “RA Application – [Your Name]” including a brief note about yourself, your availability, and your CV.
Please also fill out this Google form.

For BU students interested in RA positions:

If you are a BU student, in addition to the above, please mention whether you plan to work as part of a university program (e.g., UROP) or use this research as part of your degree requirements (e.g., thesis supervision).
For UROP, see this posting: https://www.bu.edu/urop/opportunity/responsible-ai/

Teaching. I have taught and developed courses in trustworthy AI (Data Ethics and Responsible AI, BU; Interpretability and Algorithmic Fairness, HEC Paris), analytics and machine learning (Machine Learning under an Optimization Lens, MIT; The Analytics Edge, MIT), optimization (Robust Optimization, MIT), operations management (Operations and Supply Chain Management, HEC Paris), computing (Computing in Optimization and Statistics, MIT), and introductory mathematics (Calculus I, TUC) at various levels (PhD, MS, MBA, undergraduate) and backgrounds (data science, engineering, business, business law).