Omni-Sync
Real-time collaborative Markdown editor for SRE incident playbooks. Multi-user state synchronization via Yjs-CRDTs over WebSockets eliminates the database as a single point of failure.
The Problem
During Sev-1 incidents, 20+ on-call SREs relied on static wiki pages while jumping between terminal windows — disconnected state meant duplicate debugging, stale playbook versions, and extended downtime.
System Architecture
Yjs-CRDTs over a Go WebSocket hub power zero-conflict concurrent editing across geographies. Live Prometheus metrics and fluentd logs stream directly into the editor — playbooks become the single pane of glass, not a separate tab.
System architecture diagram — coming soon
Technical Challenges & Trade-offs
Preventing CRDT state corruption during network partitions and ensuring sub-50ms propagation across regions. Built an offline-first synchronization queue that replays operations to Redis upon reconnection without conflicts.
Business Impact & Metrics
Eliminated playbook synchronization debt at the protocol level. Cut Mean Time to Acknowledge by 60% and reduced Mean Time to Resolution by 40% by embedding live telemetry directly into collaborative incident runbooks.