STACKIT AI Model Experiments is based on MLflow and enables centralized tracking of ML experiments, parameters, metrics, and models. The service complements STACKIT AI Model Serving for a complete MLOps workflow.
Features
- Experiment Tracking: Automatic logging of parameters, metrics, and artifacts
- Model Registry: Centralized management and versioning of ML models
- MLflow-compatible: Full compatibility with the MLflow ecosystem
- Integration: Seamless connectivity to STACKIT AI Model Serving and Notebooks
- Team Collaboration: Shared access to experiments and models
Typical Use Cases
MLOps Lifecycle: Data science teams track experiments, register the best models, and deploy them directly via STACKIT AI Model Serving.
Reproducibility: All experiments are stored with parameters, code versions, and dataset references, enabling results to be reproduced at any time.
Benefits
- GDPR-compliant: All model artifacts in German data centers
- Open Source Base: MLflow is an established open-source standard
- Complete MLOps Stack: In combination with Notebooks, Workflows, and AI Model Serving
- No Vendor Lock-in: MLflow artifacts are portable
Integration with innFactory
As an official STACKIT partner, innFactory supports you in building MLOps pipelines: from experiment tracking to production model deployment on sovereign STACKIT infrastructure.
Typical Use Cases
Tracking ML experiments and parameters
Model versioning and registry
MLOps workflow integration
Team collaboration on ML projects
