A decade of database platform automation

OPS Framework Claude Code force multiplication

A collaborative investigation toolkit for on-call and incident response. The engineer forms a hypothesis and Claude helps validate it: it runs commands, parses data and correlates across signals. All analysis goes through deterministic Go and Python parsers. 25 runbooks, with command audit logging and recall.

Full case study →

• Claude Code
• MCP
• Go
• Python
• YAML runbooks

Sherlock high-fidelity database forensics

Architected and deployed a framework that collects 5-second replay snapshots on 11,000+ database servers for Vitess incident response, with a DynamoDB switch hierarchy, S3 replay storage, safety-gated core dumps and an operator CLI.

Full case study →

• Go
• S3
• DynamoDB
• Chef
• Vitess

Vitess v12 → v14 upgrade vtops-go

Architected and built scalable, highly concurrent control-plane automation to upgrade Slack's 11,000+ database servers (vttablets) across Vitess versions.

• Go
• DynamoDB
• S3
• Consul

TicketD Tier-0 provisioning & DR rebar

Architected and built provisioning and disaster-recovery automation for TicketD, one of Slack's most critical services, which issues globally unique IDs to the entire application stack.

• Go
• AWS SDK
• S3
• DynamoDB
• Chef

AWS RDS database provisioner skipper

Architected, built and led adoption of the automation that fully replaced Terraform for RDS provisioning and maintenance at Auth0. Patented as US11693748B2, "Systems and methods for provisioning and decoupled maintenance of cloud-based database systems."

• Go
• AWS SDK
• S3 state
• DynamoDB metadata
• gRPC

RDS backup agent sentinel

Disaster-recovery automation that measures RTO and RPO for every RDS database across regions and environments and pages PagerDuty when targets are missed. RTO is measured by actually performing a point-in-time restore to a temporary target, which is then destroyed.

• Go
• AWS SDK
• PagerDuty SDK
• DynamoDB

Database provisioner and secrets manager fleet

Provisioned AWS RDS and MongoDB Atlas databases. Consolidated the private-cloud architecture from a dedicated database cluster per microservice to a shared topology that bin-packs environments and cuts cost.

• Go
• RDS
• MongoDB Atlas
• S3
• DynamoDB

Database migrations on Kubernetes data-migrator-cli

Orchestrated private-cloud database migrations as Kubernetes jobs, with monitoring.

• Go
• client-go
• MySQL
• MongoDB
• SQS

Fine-grained database performance monitor sentinel

UI portal on Kubernetes, a collection agent on Lambda, snapshots on S3 with metadata in DynamoDB, and the agent itself monitored with X-Ray. It was an early version of the idea that later became Sherlock.

• Kubernetes
• Lambda
• S3
• DynamoDB
• X-Ray

Bulk RDS automation saw

Fleet-wide RDS operations: SOC 2 encryption, clones, recovery, resize, reboot and parameter changes.

Database service discovery hound · sharding config pcb

Forward and reverse lookup from Kubernetes services to data clusters, driven by S3 events, Lambda and DynamoDB. Separately, a sharding configuration manager for the private-cloud database SaaS.

Open source: tabular

A Go library that simplifies printing ASCII tables from command-line utilities.