ALASCA Tech-Talk #14
A DevOps approach for AI applications at the edge
Danilo Ardagna (Associate Professor at Politecnico di Milano, Department of Electronics, Information and Bioengineering)
Artificial Intelligence (AI) is on the brink of becoming universally integrated, a development that hinges on the availability of resources at the network’s periphery. While cloud computing delivers the essential processing power for handling large datasets, edge computing plays a pivotal role: Positioned at the data generation point, edge computing is key to managing data efficiently, promptly, and securely.
This talk will delve into the innovative tools developed under the AI-SPRINT H2020 European project (https://www.ai-sprint-project.eu). AI-SPRINT has established a comprehensive framework tailored for crafting AI applications across computing continua. This framework facilitates a delicate balance between performance metrics (such as end-to-end latency and throughput) and the precision of AI models, all the while upholding stringent security and privacy standards.
Key highlights of the AI-SPRINT toolkit include:
- Simplified Programming Models: To ease the steep learning curve associated with developing AI software in computing continua, making the technology more accessible to developers.
- Specialized AI Building Blocks: Offering components for distributed training, safeguarding privacy, and employing advanced machine learning models, these components significantly reduce the time required to bring AI applications to market.
- Automated Deployment and Dynamic Reconfiguration Policies: Designed to lower the operational costs of AI software, and ensure that applications remain efficient and adaptable in the face of evolving workloads.
