These projects were built by Vitaliy Mogilevskiy as an employee of Slack and Auth0. They show the approach and the scale it's proven at; they are not HASHJOIN client engagements.
OPS Framework: Claude Code as an on-call force multiplier
An AI-native operations framework for Slack's Vitess fleet. Claude follows codified runbooks with reasoning steps and confirmation gates, every command is logged with recall, and all parsing and analysis is handed to deterministic engines. This is the pattern HASHJOIN builds for clients.
25 codified runbooks · 15 production incident analyses in 2026, each reviewed by a human
Full case study on mogilevskiy.dev →Sherlock: database forensics
Five-second snapshots of system, MySQL and vttablet state from every tablet, shipped to S3. When something breaks, you can rewind to the exact moment and see which query was burning the CPU.
11,000+ database servers · 5 s resolution vs 30 s scrapes
Full case study on mogilevskiy.dev →Vitess upgrade control plane
Concurrent control-plane automation, in Go, to upgrade Slack's vttablet fleet across Vitess versions safely and repeatably.
Full fleet roll: 35 minutes, down from more than 2 days
Experience on mogilevskiy.dev →TicketD provisioning and disaster recovery
Provisioning and disaster-recovery automation for a Tier-0 service, where downtime isn't an option.
Experience on mogilevskiy.dev →RDS provisioning and decoupled maintenance
Automation that fully replaced Terraform for RDS provisioning and maintenance at Auth0, patented as a system for provisioning and decoupled maintenance of cloud database systems.
Read the patent →Want this for your fleet?
The same approach, scoped to your stack and your incidents.