Human Data Architect

As a Human Data Architect at Surge, you'll design the systems that govern how human data is collected, structured, and delivered to train frontier AI models. This means thinking carefully about task design, evaluation methodology, and data pipeline architecture — with a constant eye on how design choices upstream affect model behavior downstream.

This role sits at the intersection of ML post-training and human data collection methodology. You'll need to understand what makes a good SFT example, what an RLHF pipeline needs to produce useful training signal, and how to structure evaluation datasets that actually discriminate between model capabilities — and then build the collection and delivery systems to produce that data reliably at scale.

• Architect end-to-end data workflows — from task design through quality assurance and delivery — ensuring each stage produces data that is well-structured for its post-training purpose (e.g., SFT, RL, model evaluation).