Cartesia is pioneering AI that interacts with the world like humans do, focusing on model architectures for training large-scale foundation models. The Human Data Operations Manager will design, scale, and operate the global evaluation workforce, impacting model quality and customer outcomes through effective workforce management and operational performance.
Responsibilities:
- Design and implement workforce structure across languages, skill tiers, and use cases, including evaluators, auditors, and leads for TTS products
- Build capacity models to support continuous eval pipelines and data production workflows
- Own relationships with vendors such as data annotation firms and contractor platforms, negotiating rate cards, SLAs, and throughput guarantees
- Decide on build, buy, or hybrid workforce models and continuously benchmark cost and performance across regions
- Design multi-layer QA systems spanning self-checks, peer review, audits, and gold tasks
- Define and track inter-rater reliability, error rates by category, and annotator-level performance distributions
- Build escalation and retraining workflows to maintain quality at scale
- Run day-to-day operations including task allocation, throughput tracking, and SLA adherence
- Build systems to reduce evaluator fatigue, rotate task types, and maintain consistency across large-scale evaluations
- Partner with tooling teams to improve evaluator UX and with data teams to ensure clean, structured outputs for model training