Abstract
In this paper, we introduce the Modular Adaptive Processing Infrastructure (MAPI), a comprehensive software suite and approach designed to streamline and enhance data analysis workflows in scientific research laboratories. MAPI selects and integrates multiple frameworks and toolkits into a web-based platform, offering a highly modular and adaptable solution for diverse data analysis requirements. By design, MAPI supports distributed processing across heterogeneous backends (edge workstations, on-premises servers, high-performance computing and public cloud), making it suitable for various beamlines and data-processing labs. This blueprint, or `recipe', provides a flexible infrastructure that can be tailored to specific experimental needs. We showcase MAPI's application through its successful implementation on the X-ray computed tomography (CT) beamline, resulting in a system for tomographic processing (STP3). The case study demonstrates MAPI's effectiveness in meeting complex computational demands, highlighting its potential for widespread adoption in scientific research environments. Most of the results reported in the paper are from a production deployment on Elettra's SYRMEP beamline using two on-premises GPU servers, but two additional ongoing deployments on different beamlines are discussed.