Knowledge of new concepts concerning data network, processing of flows and architecture focused on events becomes an increasing requirement, explains Gao. “At Sleekflow, we have built a distributed messaging infrastructure for high data flows between services and permit the possibility of performing AI models against new entrees derived from context,” he said.
“AI applications are as good as their data, but traditional data engineering approaches are insufficient for IA workloads,” adds Vaibhav Tupe, technology leader at the IT service provider Equinix and the senior IEEE member.
“Developers need skills specializing in the creation of data pipelines, creating functionality specifically optimized for automatic learning and data quality management adapted to the needs of AI,” explains TUPE. “This implies the configuration of real -time functional stores, automation of data validation and effective management of differences between training and inference data.”
