12-02, 11:35–12:05 (Europe/Amsterdam), Ernst-Curie
Building and fine-tuning models is exciting, but how do you know your model keeps performing in the way you carefully designed it? Bringing your model to production without adding any monitoring is like flying on autopilot, but blindfolded.
Adding a mature monitoring setup to your model deployments can be a daunting tasks that is often pushed off to the bottom of the to-do list, or put off entirely. How can we, Data Scientists and ML Engineers, introduce monitoring earlier in the MLOPS process and make it part of your deployment right from the start? This talk offers a practical setup to implement ML monitoring in your project using Prometheus and other open-source tools.
The ML ecosystem focuses a lot on getting models to production. However, that should not be the end goal, it’s merely the beginning of extracting real value from your model. During this talk, we will discuss:
- Why monitoring your ML model is important
- How traditional software monitoring can be used for ML systems
- What additional elements are required for ML systems
- How to recognise data drift and target drift
- Which tools are promising for ML monitoring
- A scenario for a minimal monitoring setup using open-source tools
Previous knowledge expected
I am the Lead Machine Learning Engineer at Xccelerated (part of Xebia). This means I teach and guide junior-to-medior ML Engineers in our one-year program. Besides that, I work on consultancy projects and have recently been at KLM, ProRail, and Port of Rotterdam. In my free time, I like to stay up-to-date in the ML ecosystem and play around with computer vision.