Building a Scalable Machine-Learning Service for Claims Assessment
From experiments to production.
Accompanying resources to the conference talk at Machine Learning Week Europe, 15.-16. November 2023, Berlin
Abstract
In this talk, we will delve into our journey of constructing a scalable machine-learning service for claims assessment within the automotive after-sales industry. Starting with local datasets and Jupyter notebooks, we will share how we transformed our initial experiments into a productive application. Our approach involved leveraging Azure Databricks, Kubernetes, and Seldon to enable seamless transitions from research to production, encompassing crucial aspects such as (re-)training, model deployment, and monitoring. We will also share how we transformed the starting team of data scientists into a full ML-ops application team.
Speakers
- Ole Brodersen: ole_bjoern.brodersen@, LinkedIn, Job Portrait at heise.de
- Thomas Hummel: thomas.h.hummel@, LinkedIn
