projects

Open-source on-call alert management. Built as a self-hosted alternative to PagerDuty and Opsgenie — the kind of tooling I've worked with professionally for years, rebuilt from first principles with a complete feature set.

Ingests alerts from Prometheus Alertmanager, Slack, and generic webhooks. Routes them through content-based rules to multi-step escalation policies with configurable delays and ack timeouts — preventing the classic "ack and abandon" problem. On-call schedules support daily, weekly, and custom rotations with overrides. Notifications go out via Slack (with interactive ack/resolve buttons), SMS, voice, email, and mobile push. Includes maintenance windows that suppress escalation without losing audit trail, priority-based routing so P4/P5 alerts don't page anyone at 3am, and time-window grouping to prevent notification storms.

Ships with a React Native mobile app with critical alerts that bypass Do Not Disturb, a Terraform provider for managing teams and policies as code, an MCP server for AI assistant integration, Kubernetes Helm charts, SSO via OIDC, and analytics covering MTTA, MTTR, and per-engineer load distribution.