The growing importance of sustainability perceived by governments and the public puts pressure on organizations to adapt their value-creating processes. I study how organizations can optimally redesign their operations and their business models in order to make effective use of emerging technologies and better align profitability objectives with social and environmental sustainability requirements. Guided by data and insights from practice, I examine organizations within their broader supply chain context.


Here you can find abstracts of my papers.

Modern traceability technologies promise to improve supply chain management by simplifying recalls, increasing visibility, or verifying sustainable supplier practices. Initiatives developing traceability technologies — and who hope to make their technologies the industry standard — must choose the least-costly set of firms to target as early adopters. This choice is challenging because firms are part of supply chains interlinked in complex networks, yielding an inherent supply chain effect: benefits obtained from traceability are conditional on technology adoption by a (potentially large) subset of firms in a product's supply chain.
We prove that the problem of selecting the least-costly set of early adopters in a supply chain network is hard to solve and even approximate within a polylogarithmic factor. Nevertheless, we provide a novel linear programming-based algorithm to identify the least-costly set of early adopters. The algorithm is fixed-parameter tractable in the supply chain network's treewidth, a parameter which we show to be low in real-world supply chain networks. The algorithm also enables us to derive easily-computable bounds on the optimal cost of selecting early adopters as well as key managerial insights about which type of firm to select.

With Andre Calmon (Georgia Tech) and Georgina Hall (INSEAD), available on SSRN (currently undergoing revision).
Invited for major revision (second round) at Management Science.
Family planning services play a central role in the sustainable development agenda by improving health and education while reducing poverty and gender inequality. Marie Stopes International (MSI) reaches underserved clients in rural areas with the help of mobile healthcare units (MHUs). This approach is gaining traction more generally to enable broad access to health services for marginalized rural populations, thus stepping closer to fulfilling SDG 3: Good Health and Well-Being. Our research addresses the complex challenge of allocating limited mobile healthcare unit resources to create the largest possible impact in the long-term, especially in light of intricate demand dynamics.
We model the MHU resource allocation problem as the optimization of a sum of sigmoidal functions and derive new theoretical results with operationalizable managerial insights. We further develop a machine learning approach to predict demand in such a setting that provides interpretable results, a key requirement in the humanitarian context. Our empirical application combines MSI's client demand data with rich, publicly available data sources. We use the resulting estimates to demonstrate the benefits of our analytically derived insights.

With Andres Alban (Harvard University), Harwin de Vries (Erasmus University), & Luk Van Wassenhove (INSEAD).
Published in Manufacturing & Service Operations Management, 2022, Vol. 24, Issue 6.
Finalist, 2022 MSOM Society Award for Responsible Research in Operations Management.
Technological advances enable new business models for heavy equipment manufacturers, replacing or augmenting ownership-based models with access-based consumption. An ever-increasing list of peer-to-peer sharing platforms (e.g., Trringo by Mahindra, Yard Club by Caterpillar, or Coop by Ryder) shows that it is paramount for decision-makers to understand the economic implications of such business models. However, the efficacy of emerging models relying on access-based consumption is unclear. A lack of empirical evidence of their performance motivates the need for analytical insights. This paper focuses on understanding the performance of different emerging business models, particularly that of peer-to-peer product sharing, by considering salient economic and operational factors in the context of heavy equipment.
Although sharing business models are widely publicized, the optimal model for a heavy equipment manufacturer will depend on operational factors linked to after-sales services, a point which has been previously overlooked and deserves closer attention by decision-makers. We show when a heavy equipment manufacturer prefers setting up a sharing platform for its products, which provides a positive rationale for the emergence of such manufacturer-owned platforms in practice. We also describe the optimal design of a sharing platform, highlighting the need to subsidize one part of the business (sharing) to the benefit of another (after-sales services). Moreover, we provide insights into how manufacturers can leverage after-sales services when threatened by a third-party sharing platform's entrance.

With Niyazi Taneri (Cambridge University) & Sameer Hasija (INSEAD), available on SSRN (currently undergoing revision).
Invited for major revision at Operations Research.

Selected Works in Progress

Here you can find a description of some of my other research project.

One of the key drivers of modern supply chain traceability solutions lies in the underlying information systems. However, it is unclear how centralized cloud-based systems and decentralized blockchain systems—both modern solutions that are believed to drive a wave of improved traceability adoption—compare to traditional ERP-powered traceability and each other.
We build a game-theoretic framework to show that differences between information systems profoundly impact both the adoption and effectiveness of traceability solutions. In particular, modern traceability systems allow for a different mode of connection (1:n instead of 1:1), and tend to be cheaper and quicker to adopt. They differ, moreover, in their approach to data interchange and ownership. We show that all these differences can have non-intuitive impacts on whether traceability is broadly adopted because they also affect long-term supplier relationships and commitments. For example, the degree of competition in a supply chain network largely determines the advantage of modern over traditional traceability systems. This is because modern systems' efficiency advantages coincide with issues in sustaining long-term supplier relationships. We also show that retention of data ownership, one of the key arguments for blockchain technology, may further compromise supply chain members' ability to commit to data sharing. This points to a potential pitfall in the usage of blockchain technology for traceability.

Joint work with Andre Calmon (Georgia Tech) & Sameer Hasija (INSEAD).

I discussed some of my research results in a tech talk hosted by digital@INSEAD. Together with my Ph.D. advisor Sameer Hasija, my colleague Dmitry Sumkin, and Aly Madhavji from the Blockchain Founders Fund, we analyzed different perspectives and distinguish the hype from reality.
Supply chain models often remain untested beyond case studies because real-world data is costly and has several limitations. To circumvent these, we present the first random generative model of supply chain networks that enables a simple process for creating networks closely resembling those observed in practice.
We can bound the generated networks' treewidth by a function of the input parameters. This is important because many optimization problems which are hard to solve over general graphs admit a fixed-parameter tractable algorithm in the treewidth of the graph. We illustrate this point with the NP-hard guaranteed service model, showing that the problem can be solved in pseudo-polynomial time even on networks that do not form trees but have treewidth logarithmic in the number of firms (as is the case for the networks generated via our model).

Joint work with Andre Calmon (Georgia Tech) & Georgina Hall (INSEAD).
Many everyday tasks require sequential, interlinked decisions that make it difficult for workers to estimate the best strategy. As a result, Artificial Intelligence (AI)-based recommendation systems are applied increasingly to simplify complex decision-making processes. However, due to human biases, there remains a gap between what is technically possible and what is utilized. This work seeks to understand how recommendation systems can help induce efficient behavior in such tasks while considering human biases.
We develop a sequential decision-making task in the form of a virtual electric vehicle driving game in which the participant, equipped with algorithm-generated recommendations, needs to make sequential charging decisions while facing uncertain traffic. Our experimental results offer key insights into how humans make decisions and respond to those recommendations. Based on these insights, we develop a recommendation system building on deep reinforcement learning. This allows us to simultaneously learn drivers' actions for a given recommendation and the optimal recommendation to suggest in consequence.

Joint work with Park Sinchaisri (University of California, Berkeley).
Many resource allocation decisions are not based purely on utilitarian considerations but capture some notion of fairness. Technological developments in areas such as personalized medicine or machine learning have made such questions of fairness more pressing. At the same time, allocation decisions tend to have inherently uncertain impacts in these areas. Operations management has provided a deep understanding of the trade-off between fairness and efficiency but has often ignored the inherent uncertainty. We demonstrate that ignoring uncertainty in planning can systematically bias decisions away from already disadvantaged agents. Moreover, we show how fairness problems from different domains can be analyzed systematically from a unifying optimization perspective.

Joint work with Peter Zhang (Carnegie Mellon University).