Sayiir Server

Sayiir Server is under active development. Join the Discord for early access and to shape the roadmap.

Sayiir’s open-source core gives you a fast, simple durable workflow library. Sayiir Server adds the operational layer — a managed platform for teams that need visibility, scheduling, multi-tenancy, and production-grade security without building it themselves.

Same workflows. Same code. Just deploy to the server instead of running standalone.

Planned Features

Web Dashboard

Visual workflow monitoring with real-time execution status, task-level drill-down, execution history, and searchable workflow instances.

Multi-Tenant Isolation

Namespace-based tenant isolation with per-tenant resource quotas, role-based access control (RBAC), and API key management.

Managed Scheduling

Cron-based triggers, event-driven workflow starts, webhook ingestion endpoints, and recurring workflow schedules — all configurable from the dashboard.

Observability

Native OpenTelemetry traces, Prometheus metrics endpoints, structured logging, and configurable alerting rules for workflow failures and SLA breaches.

Security & Encryption

Encryption at rest — Workflow state, checkpoints, and signal payloads encrypted with AES-256-GCM
Encryption in transit — Mutual TLS (mTLS) between workers, server, and storage backends
Secret management — Inject secrets into tasks without persisting them in workflow state
Audit logging — Tamper-evident audit trail for all workflow operations (start, pause, cancel, signal, admin actions)

Kubernetes-Native Deployment

Helm chart — Single helm install sayiir for the full stack (server, dashboard, workers, PostgreSQL)
Worker affinity / anti-affinity — Schedule workers on specific nodes based on labels, taints, or topology constraints. Ensure GPU tasks land on GPU nodes, keep memory-heavy workers spread across zones
Horizontal Pod Autoscaler — Auto-scale workers based on task queue depth metrics
Rolling upgrades — Zero-downtime workflow deployments with graceful worker drain

Time-Critical Task Scheduling

Priority queues — Assign priority levels to workflows and tasks. High-priority tasks are claimed before low-priority ones, even under load
Deadline-aware scheduling — Tasks with approaching deadlines are automatically promoted in the queue
Preemption — Optional preemptive scheduling where time-critical tasks can interrupt lower-priority work
SLA monitoring — Define SLA targets per workflow. The dashboard tracks compliance and alerts on breaches before they happen

Architecture

Sayiir Server runs as a lightweight gRPC service alongside your existing PostgreSQL backend. Workers connect to the server instead of polling the database directly, enabling centralized scheduling, authentication, and observability.

Your App → Sayiir Server (gRPC) → PostgreSQL
               ↕
         Worker Pool (auto-scaled)
               ↕
         Dashboard (web UI)

Your workflow code doesn’t change. The same #[task] functions, the same Flow builder, the same workflow! macro — just pointed at the server endpoint instead of a direct database connection.

Get Involved

Sayiir Server is being designed in the open. We’re looking for:

Design partners — Teams with production workflow needs who want to shape the feature set
Early adopters — Try pre-release builds and provide feedback
Contributors — Help build the server, dashboard, or Helm charts

Join the Discord or open an issue to get started.