A Systemic Approach to Machine Learning Project Management

Despite their increasing popularity, artificial intelligence (AI) and machine learning (ML) projects often fail to deliver expected outcomes, frequently ending prematurely or not being deployed. Beyond data and model requirements, theseprojects necessitate cross-domain performance, business analysis, effective stakeholder and team management, risk and quality management, and a tailored development approach. These factors underline the need for a systemic project management (PM) approach that addresses the interconnections among business, organizational, team, and PM factors, rather than focusing on isolated achievements of AI and ML. Current PM methods such as Cross-Industry Standard Process for Data Mining, Team Data Science Process, Scrum, and Kanban may not adequately meet these challenges. Therefore, a systemic PM framework is essential for artificial intelligence and machine learningproject success, particularly in critical areas such as healthcare. This article argues for a change toward a principle-based, holistic, and systemic PM approach tailored to the performance domains of artificial intelligence and machine learningprojects. It explores the adoption of the Project Management Body of Knowledge (PMBOK V7) for an ML project at Baskent University Hospital Ankara. By combining systems engineering and Soft Systems Methodology with a survey on performance domains, we enhance PMBOK V7’s applicability, proposing solutions that include conceptual and process models. Findings indicate that PMBOK V7 meets the necessary project requirements but needs adaptations for ML processes.