AIZOYA

AIZOYA’s Engineering That Remembers helps coding assistants resume engineering work from verified evidence instead of starting over. MongoDB Atlas preserves the mission contract, sessions, failed checks, decisions, candidate hashes, and independent verification results. Our demonstration begins with a transaction importer passing 9 of 10 checks. In a second session, the operator retrieves the recorded Unicode parsing failure from Atlas and supplies it to Kiro. Kiro repairs the parser, and independent verification passes all 10 checks. The dashboard shows the failure, retrieved evidence, repair decision, and verified outcome. The project combines a Python engineering harness with a TypeScript/React presenter deployed through Modelence. This is a demonstrated human-assisted recovery workflow using fictional data, with MongoDB Atlas providing persistent memory and traceable evidence across sessions.