Zepey
Description
Zepey started life in 2006 as an online file-storage site. The name has been reborn as a modern, multi-tenant project management platform that I designed and built from the ground up with AI-assisted development. It now runs the day-to-day engineering work at Mod Advisor, where it replaced Asana company-wide.
Projects get kanban boards, task lists, sprints, a timeline with drag-to-reschedule, collaborative whiteboards, decision logs, file attachments, and a QA and DevOps workflow that walks each task through sandbox, stage, and production with required testers and approvals. Pull requests from GitHub, GitLab, Bitbucket, and Azure DevOps link straight to tasks, and Slack, Teams, and Asana integrations keep notifications and imports flowing. A provider-neutral AI layer (Anthropic, OpenAI, Gemini, Grok, or a self-hosted Ollama model) improves task descriptions, estimates sprints, and powers a documentation chat, and Zepey exposes its own MCP server so AI agents can work inside it. Organizations sign up with per-seat Stripe billing, and security includes TOTP two-factor auth, row-level tenant isolation in Postgres, and per-organization KMS keys for private files in S3.
The infrastructure is deliberately lean. Three EC2 instances in AWS sit behind Cloudflare and Caddy: a Dockerized Next.js app, a dedicated realtime server for live whiteboard collaboration and update streaming, and a self-managed Postgres 16 instance that I chose over RDS at roughly half the cost. My own backup script dumps the database every 12 hours and copies it to a second AWS region. Deploys are a single script that pulls, builds, migrates, health-checks, and rolls back automatically on failure.