Infrastructure Review
The Data Science Without Borders (DSWB) project infrastructure is designed to uphold the autonomy, data protection, and security protocols of each participating country (pathfinders). A hybrid implementation has been identified as the most suitable approach, leveraging cloud resources for specific computational tasks while keeping data storage and general computations localized. The project prioritizes utilizing each country’s existing infrastructure, which is predominantly on-premise. Key tools necessary for the project are securely deployed on local servers, with controlled access mechanisms in place to facilitate collaboration and support within the project.
The project spans three countries-Ethiopia, Cameroon, and Senegal-each hosting a dedicated research institution. Since the level of digital technology adoption varies across these countries, the project approach is customized to address the unique needs and contexts of each location, including additional training.
During an initial country-specific visit, discussions focused on identifying synergies between the pathfinders’ data teams and the project’s technical working group. These discussions explored optimal ways to utilize existing resources and deploy additional ones based on each pathfinder’s infrastructure desk reviews. A key outcome was the need for technical upskilling in deploying open-source technologies to support data standardization and leveraging cloud technologies, both of which are critical components of the project. (DSWB Pathfinders Infrastructure Desk Review report)
DSWB is developing two platforms: a no-code AI tool and a data harmonization and analytics pipeline. Both platforms will be deployed on-premises using Docker technology on dedicated servers. The pathfinders will maintain these deployments to enable continuous data standardization and facilitate the use of machine learning to answer the country’s specific research questions. To promote research collaboration, the project will consider a central cloud deployment that supports durability, availability, and scalability-key features of a well- architected framework (WAF).
