Can AI replace Crowdin?

Build a personal replacement for Crowdin in an empty repository.
Use Next.js 15, TypeScript, PostgreSQL, Drizzle ORM, and one optional machine-translation API; do not offer alternative stacks.
The core loop is: manage a software localization project's translation keys with file sync, a glossary, and reviews, preserve context, import and export standard files, and publish approved strings.
Make the first run work locally with one documented command.
Store all user data locally by default and make export straightforward.
Put secrets in .env, ship .env.example, and never commit credentials.
Implement projects, locales, keys, namespaces, descriptions, screenshots, translations, and review status.
Import and export JSON, YAML, PO, and XLIFF while preserving placeholders and plural forms.
Add glossary checks, missing-string filters, translation memory, comments, and reviewer assignment.
Use machine translation only on selected strings and show provider, cost, and source text before approval.
Provide token-authenticated pull and push endpoints for CI with an immutable change log.
Add backups, locale deletion safeguards, and a pseudo-localization preview.
Include clear empty, loading, success, and recoverable error states.
Add input validation, safe filenames, and graceful handling of unavailable APIs.
Write focused tests for the core transformation and one end-to-end happy path.
Create a README with setup, architecture, permissions, data location, and backup steps.
Do not add accounts, billing, telemetry, analytics, or a hosted control plane.
Deliberately leave out a translator labor marketplace.
Deliberately leave out a global website translation proxy.
Deliberately leave out enterprise integrations, legal translation assurance, and round-the-clock support.
Finish by running the tests and listing the exact commands used.