Machine learning presents a new frontier in security challenges for organisations. In this session, we’ll cover the combination of ML infrastructure, Developer operations and Security policies that must be implemented to tackle this problem.

Deploying and maintaining machine learning systems has uncovered new challenges, particularly when running at scale and in production. These systems require fundamentally different approaches to the traditional software and DevOps spaces.

In this talk, Adrian Gonzalez-Martin, Machine Learning Engineer at Seldon, will outline the field of security in data and ML infrastructure including the key challenges and opportunities it presents. He’ll dive into a number of practical examples and the reasoning behind the eight LFAI ‘Principles for Trusted AI’.

He’ll showcase how to leverage cloud-native tooling to mitigate critical security vulnerabilities and will cover essential concepts such as:

– Role-based access control for ML system artifacts and resources

– Encryption and access restrictions of data in transit and at rest

– Best practices for supply chain vulnerability mitigation

– Tools for vulnerability scans

– Templates that practitioners can introduce to ensure best practices