What should I use for a developer-friendly pipeline framework?

If you want a developer-friendly pipeline framework, I’d usually suggest choosing based on what kind of pipelines you mean:

Good general-purpose options

• Dagster — probably the best pick if you want a modern, developer-friendly experience.

  • Strong typing/assets model
  • Great local dev workflow
  • Good observability and testing
  • Nice if you care about maintainability

• Prefect — very approachable and flexible.

  • Easy to get started
  • Great for Python-first teams
  • Good orchestration without too much ceremony
  • Nice when you want simplicity and fast iteration

• Airflow — still common, but less “developer-friendly” than the two above.

  • Huge ecosystem
  • Mature and widely used
  • Can feel heavy/verbose for new pipelines

If you mean data/ML pipelines specifically

• Dagster for data engineering workflows
• Prefect for simpler orchestration and ops-heavy workflows
• Kubeflow if you’re deep into Kubernetes-based ML pipelines, though it’s not the easiest to work with

My practical recommendation

• Choose Dagster if you want the best developer experience overall.
• Choose Prefect if you want the fastest path to something working.
• Choose Airflow only if your team already uses it or you need its ecosystem.

If you want, I can also give you a comparison table for Dagster vs Prefect vs Airflow based on team size, complexity, and deployment style.