// the find
kestra-io/kestra
Event Driven Orchestration & Scheduling Platform for Mission Critical Applications
Kestra is a YAML-based workflow orchestrator that competes with Airflow, Prefect, and Temporal. It handles both scheduled and event-driven pipelines with a built-in UI, 500+ plugins, and supports running tasks in any language via Docker/K8s. The target audience is data engineers and platform teams who want something less Python-centric than Airflow but more opinionated than raw K8s CronJobs.
The declarative YAML model is actually clean — tasks, triggers, and error handling are all first-class in the schema, not bolted on. The UI-to-code sync is a genuine differentiator: edits in the visual editor write back to YAML, so you don't end up with two sources of truth. The plugin ecosystem (Kafka, S3, BigQuery, Spark, HTTP, SQL, etc.) is broad enough that most pipelines don't require custom code. Task runners with Docker/K8s isolation mean you can run Python 3.9 in one task and Go in the next without VM-level overhead.
The open-source tier quietly stops short of production essentials — RBAC, SSO, secrets management, and multi-tenancy are all Enterprise Edition. You won't know this until you're building something real and hit the paywall. The local mode runs everything in a single JVM with embedded H2/internal queues, so the dev experience doesn't reflect how the distributed deployment (separate executor, scheduler, worker, indexer processes) actually behaves — bugs in distributed state handling won't surface locally. Java as the extension language for plugins is a high barrier for teams that live in Python. The Helm chart values.yaml is sprawling and poorly documented for non-trivial deployments.