


LakeDB
Built and shipped a cross-platform SQL client with multi-database workflows and review-first AI for SQL generation and query correction.
- The context
- Database workflows often get fragmented across different clients, connections and engines. I wanted one workspace that could keep connection context isolated, support real database operations, and add AI assistance without giving an agent permission to execute SQL automatically.
- What was shipped
- I built LakeDB as a desktop SQL client for MySQL, MariaDB and SQLite. It includes independent workspaces per connection, schema-aware SQL editing, table browsing and editing, database comparison and reviewable migrations, production safeguards, and QuerIA for context-aware SQL generation and AI-assisted query correction.
- Operator contribution
- I designed and built LakeDB mostly solo, including the desktop architecture, database integrations, SQL workflows, QuerIA integration, backend/API communication, privacy and execution boundaries, packaging/releases, and an enterprise deployment model where companies can control their own AI configuration and API infrastructure.