@article{pahwa_starly_2021, title={Dynamic matching with deep reinforcement learning for a two-sided Manufacturing-as-a-Service (MaaS) marketplace}, volume={29}, ISSN={["2213-8463"]}, DOI={10.1016/j.mfglet.2021.05.005}, abstractNote={Suppliers registered within a manufacturing-as-a-service (MaaS) marketplace require near real time decision making to accept or reject orders received on the platform. Myopic decision-making such as a first come, first serve method in this dynamic and stochastic environment can lead to suboptimal revenue generation. In this paper, this sequential decision making problem is formulated as a Markov Decision Process and solved using deep reinforcement learning (DRL). Empirical simulations demonstrate that DRL has considerably better performance compared to four baselines. This early work demonstrates a learning approach for near real-time decision making for suppliers participating in a MaaS marketplace.}, journal={MANUFACTURING LETTERS}, author={Pahwa, Deepak and Starly, Binil}, year={2021}, month={Aug}, pages={11–14} }
 @article{pahwa_starly_cohen_2018, title={Reverse auction mechanism design for the acquisition of prototyping services in a manufacturing-as-a-service marketplace}, volume={48}, ISSN={["1878-6642"]}, DOI={10.1016/j.jmsy.2018.05.005}, abstractNote={The affordability and increased capability of additive manufacturing machines has spawned prototyping service bureaus throughout the world. This poses a challenge to designers who are looking to obtain quality 3D printed parts at best available prices within fastest turnaround times. Customers relying on a sole source for 3D printed parts may have limited options in obtaining the best deals. From a service supplier point of view, filling excess capacity will require significant marketing budgets to reach and retain customers. In this paper, we present a novel mechanism design approach for improving the accessibility of prototyping services providers by leveraging their excess capacity. In our proposed mechanism, consumers name their own price and the mechanism will find service bureaus who are willing to make the part under the stated price. The mechanism runs similar to a reverse auction where consumers bid and the platform finds a service supplier which is able to match the stated bid price. The incentive for suppliers to participate in such a platform is the opportunity to market their excess capacity to a deal conscious consumer at a lower price without cannibalizing their existing sales channels. Qualified suppliers do not directly compete with each other for any given order since they are chosen using a two stage selection process by the service platform. This algorithm ensures that every supplier has a fair chance of selling its services on the platform regardless of price. We implement the proposed mechanism design approach in a simulated service marketplace and empirically evaluate the marketplace behavior by studying the impact of various model factors such as the supplier threshold price, the size and variety of suppliers in the marketplace.}, journal={JOURNAL OF MANUFACTURING SYSTEMS}, author={Pahwa, Deepak and Starly, Binil and Cohen, Paul}, year={2018}, month={Jul}, pages={134–143} }