Скачать или смотреть Training and Ensuring Reliability of ML Models at Wolt: The Power of ArgoCD, Flyt... Stephen Batifol

Training and Ensuring Reliability of ML Models at Wolt: The Power of ArgoCD, Flyte, and Argo Workflows - Stephen Batifol, Wolt

In recent years, Machine Learning models have become ubiquitous in many industries. However, ensuring the reliability and robustness of these models can be a challenge. Significant demand for compute resources and difficulties to ensure consistency during training and deployment phases, make it harder to achieve the goal of more scalable and reliable models. In this talk, we will delve into the practices that Wolt, a technology company known for its local commerce platform, has implemented to train and maintain their ML models. We will cover the use of ArgoCD, Flyte, and Argo Workflows to provide a streamlined and automated process for training, testing, and deployment while ensuring the reliability of ML models. Additionally, we will discuss how these tools provide a unified platform for collaboration between Data Scientists and ML teams. By the end of this talk, attendees will have a better understanding of the importance of ensuring reliability of ML models and the various tools that can be used to achieve this goal