The observability of systems equipped with artificial intelligence is now a key piece in ensuring their proper functioning and operational performance and also for achieving reproducibility of experiments and checking that they are working correctly when they have an impact on the general public. Feature stores, model stores and evaluation stores as specific purpose systems make it possible to identify data drift, unwanted behaviour or potential risks in these systems so that they can be remedied as quickly as possible.
In this session we will set out the key pieces for observability of AI systems in an MLOps framework, their functionality and use in the everyday operations of organisations.
Addressed to:
- Data platform and data science managers
- Solution architects in advanced analytics
- Functional architects
- Machine learning engineers
- Data scientists
- Data engineers