Case Study: Spendesk automates ML pipelines at scale with Valohai

Automating machine learning pipelines for spend management

Spendesk, a FinTech SaaS company, needed a more robust way to manage end-to-end machine learning pipelines for its spend management platform, including payment auto-categorization and other internal ML use cases. The team was using Airflow and MLflow, but needed a single, secure, all-in-one platform that could handle weekly retraining, production deployment, and sensitive financial data at scale.

With Valohai, Spendesk standardized ML pipeline automation across training and deployment in one workflow, enabling the team to pre-process data, train models, aggregate outputs, deploy to production, and revoke old versions automatically. Valohai helped Spendesk train more than 3,600 models in parallel every week and complete over 250,000 training runs per year, while reducing dependency on DevOps/IT and improving team autonomy and experimentation speed.

Spendesk

Adèle Guillet

Senior Lead Data Scientist