Data Science Roadmapping for AI Alignment: Insights From a Multivocal Literature Review and a Retrospective Case Study

Many organizations struggle to keep their artificial intelligence (AI) systems aligned with operational data and computing resources in today's volatile landscape. Data Science Roadmapping (DSR) embeds data layers into planning scenarios and enables a human-centric process. While DSR is effective for creating data science roadmaps, it lacks a clear implementation framework. This research advances DSR as a continuous AI alignment platform through three phases: (1) a multivocal literature review (MLR) of academic and grey sources identifies gaps and tools; (2) synthesis of these findings adapts DSR for ongoing AI alignment; (3) a retrospective case study evaluates the adapted process. Initial results show the effectiveness of agile modifications to the DSR framework and the integration of a real-time platform for roadmap implementation and monitoring. Case study participants strongly supported a dedicated roadmapping operations team, especially to manage communication, detect AI deviations, and ensure compliance. This underscores how the operationalization of roadmapping can strengthen data and AI governance.