[OSS GAP] Annotation / Labeling #ai #labeling #dataset #workflow
Pain: Small teams building datasets need straightforward queues and review, not an enterprise annotation platform with procurement-level pricing. “TLDR: The old maintainer appears to have sold the extension to parties unknown, who have malicious intent to exploit the users of this extension in advertising fraud, t” — github-issues (https://github.com/greatsuspender/thegreatsuspender/issues/1263)
Why now: Small teams in 2026 are cutting tool spend and refusing extra platform debt. Labelbox, Scale, Label Studio Enterprise are strong products, but they are packaged for bigger companies than ML engineer and other small teams. That makes a smaller, self-hosted wedge in annotation / labeling unusually easy to explain.
Tiny wedge: Label queues for ML engineer and other small teams without Labelbox-style pricing and platform weight.
Why this wins: Replaces recurring Labelbox spend with a boring self-hosted alternative for ML engineer and other small teams.
Scope cut: Skip managed workforce and RLHF platform in v1.
Stack: Node.js + Express + PostgreSQL + Redis + background worker + S3-compatible object storage + Provider adapters.