MLOps Community Amsterdam Meetup @ Picnic - Model Observability

Welcome to a new live get together for the global MLOps.community in Amsterdam. Together with our host, Picnic, we will enjoy a series of talks and ample time to socialize with others in the community!

Schedule
18.00-18.30: Walk in
18.30-19.00: Model Observability Using Picnic's Machine Learning Platform - Tom Steenbergen and Job Almekinders
19.00-19:30: Break + drinks and bites
19.30-20:00: Towards Understanding the Behavior of Deployed Models Over Time: A Study Case on AIOps - Lorena Poenaru-Olaru
20:00-20:30: Lighting talks (sign-up first come first serve)
20.30-21.30: Networking + drinks and bites

Sign-up instructions:

• Sign up in the meetup page
• *NEW*: Get your free ticket via Lu.ma:[ https://lu.ma/6qutunp2]( https://lu.ma/6qutunp2)
• Let us know if you have any strict dietary restrictions (e.g. vegan🌱)
• We are looking for speakers for the next events. If you would like to give a talk, let us know the topic and a contact information.

🎤 Talks

Talk 1:
Model observability using Picnic's Machine Learning Platform
At Picnic Technologies, we’re revolutionizing the way people buy groceries. Our affordable and sustainable service is made possible by cutting-edge technology and passionate engineers. Machine learning is at the heart of many of our operations. From time-series forecasting for the supply chain to recommender systems for our customer app.

In this presentation, Tom Steenbergen, ML Platform Lead, and Job Almekinders, MLOps Engineer, will dive into Picnic’s machine learning platform and how we enable ML Engineers to easily monitor their ML models.

First, Job Almekinders will share more about Picnic's journey of building an ML Platform and how it now empowers ML engineers to easily run tens of models in production. Following the platform overview, Tom Steenbergen will take the stage to discuss in more detail a crucial standardized solution of the platform to ensure all predictions are logged to Snowflake to improve model governance and enable automatic model monitoring at Picnic.

Talk 2:
Towards Understanding the Behavior of Deployed Models Over Time: A Study Case on AIOp
Speaker: Lorena Poenaru-Olaru

The evolving character of real-world data can significantly impact the performance of AI/ML systems over time. Therefore, AI/ML systems need to be constantly monitored against data changes to ensure their accuracy is not degrading. “Which technique shall I use to monitor my model?” remains still an open question. In this talk I will highlight different model monitoring techniques for both univariate (time series) and multivariate data and explain how we chose the most suitable model degradation indicators for two AIOps applications (disk and job failure prediction).