Running Distributed AI Workflows with ClearML

Discover how modern AI teams move beyond isolated notebooks and build reproducible, distributed machine learning workflows with Dario Cavarretta and Barbaros Özdemir.

In this 90-minute IBM Developer Austria Meetup session, we will explore ClearML through a practical, live, demo-driven format. Rather than focusing only on slides and theory, we will demonstrate how experiments are created, tracked, executed remotely, and managed as part of a complete MLOps workflow.
You will see how ClearML can help teams:

• run AI workloads on remote machines while controlling them from a laptop;
• track experiments, logs, metrics, parameters, and artifacts;
• manage and version datasets and models;
• build automated and reproducible machine learning pipelines;
• compare model performance across multiple runs;
• recover from failed data-preparation steps;
• keep data and model execution on controlled infrastructure.

Using the TACO dataset for litter detection, we will demonstrate an accessible environmental use case while keeping the main focus on architecture, distributed execution, and the practical value of MLOps.

## Speakers

Barbaros Özdemir — Host and introductory speaker
Senior Managing Consultant and Level 2 Architect at IBM. Barbaros will open the event and provide a practical introduction to ClearML, its architecture, and its core workflow.
Dario Cavarretta — ClearML expert and main speaker
Data Scientist at IBM specializing in visual recognition. Dario will lead the advanced technical demonstration and discuss challenges he encountered while using ClearML in a real client project, along with the solutions and improvements he developed to address them.