Funded by: ServiceTitan
PI: Habet Madoyan, Aram Butavian
Researchers: Tigran Gaplanyan, Davit Badalyan, Gurgen Kolotyan, Hayk Daghunts, Kristine Hambardzumyan, Laura Barseghyan
Â
Project description
The research initiative, led by Habet Madoyan (PI) and Aram Butavyan (research lead), brought together students and alumni from the American University of Armenia, along with applicants from other universities, to work on industry-driven machine learning problems in collaboration with ServiceTitan. In total, fourteen participants were involved across three projects, each lasting seven months and combining a detailed review of the literature with the development of practical solutions, with the aim of producing academic publications.
The first project focused on identifying duplicate materials across large-scale product catalogs, addressing challenges arising from millions of entries distributed across thousands of catalogs. The research aimed to develop machine learning approaches capable of accurately matching product descriptions while maintaining scalability and efficiency. The second project built on historical service data and investigated whether both structured and unstructured information could be used to simulate a conversational process for scoping new jobs, generating clarifying questions, retrieving similar past cases, and producing accurate price estimates through a sequence of machine learning models. The third project is examining the process of estimate and proposal creation in on-site sales settings, exploring how data-driven methods can assist technicians in generating structured and effective pricing options while accounting for practical constraints such as material availability and service logistics.
The initiative provides a structured pathway for students to engage in applied research through a competitive selection process, where applicants are required to solve a task directly related to the project topic.
Throughout the projects, students work with real-world datasets and industry-relevant challenges, gaining hands-on experience that goes beyond traditional coursework. They are involved in all stages of the research process, including literature review, model development, evaluation, and interpretation of results. This exposure allows them to develop both practical technical skills and a deeper understanding of how machine learning methods are applied in realistic settings.
The experience also strengthens participants’ ability to think critically, communicate technical ideas, and collaborate within a research team. By working on tasks that closely reflect industry needs, students gain valuable insights into professional workflows and expectations, which supports their preparation for future careers in data science and machine learning. At the same time, the initiative fosters meaningful interaction between students, researchers, and industry practitioners, reinforcing the connection between academic training and applied research.