Recapping on Software and Data Management

• Giada Malatesta (Istituto Nazionale di Fisica Nucleare)

Room: Sala Venturi In a highly heterogeneous and distributed environment, the adoption of solutions for sharing software, libraries and their configuration files is a necessity for the scientific community. The Software Management Service has been created to make sharing efficient, user-friendly, and usable in multidisciplinary environments, aiming to achieve two fundamental aspects: automation and abstraction. The exponential increase in the amount of data produced is making data management a key issue. In this context, the DataLake model was born, in which geographically distributed physical storages are federated, providing a level of abstraction that allows the user to interact with data transparently. The aim of implementing this model was to study it's application in other communities, such as life science, using Data Management tools (RUCIO and FTS) that had already been used in some large experiments.

Summary on activities: Blockchain for Data Integrity, Quantum Optimization, and AI Middleware

• Domingo Ranieri (Istituto Nazionale di Fisica Nucleare)

Room: Sala Venturi In this seminar, I will present an overview of my recent activities across three projects. First, as part of Interoperable Data Lake Innovation Grant, we contributed to the instantiation of a local blockchain using Hyperledger Fabric to enhance the validation of data and metadata stored in the Rucio Datalake by leveraging cryptographic hashes recorded on the blockchain. Second, in collaboration with the University of Bologna, we are exploring quantum computing approaches to optimize biomechanical muscle function, investigating novel solutions for complex optimization problems. Lastly, I contributed to the AI-Ranker component within the Orchestrator, improving decision-making processes through artificial intelligence techniques. The activity primarily focuses on middleware development, involving the creation of two components that consume Kafka messages and utilize the extracted data to train and infer Machine Learning models, which are then stored in MLflow.