Scale AI

A current example: Technical Advisor Specialist (Part-Time Internship), remote within the US, summer-semester scheduling at 5–20 hours per week, $50/hour (listing).

Scale's early-career footprint is small and req-driven rather than cohort-driven, so the practical implication is that the relevant deadline for a given role is set by the individual posting, not by a published season – a candidate hunting at Scale is hunting for live reqs, not preparing for an annual application window.

First, Scale's stated mission is to "develop reliable AI systems for the world's most important decisions," and its product surface bifurcates clearly into Data at Scale (training data, annotations, RLHF for foundation-model developers), Evaluations (model benchmarking and red-teaming), and Applied AI (full-stack AI for enterprise and government) – the data-infrastructure layer beneath the foundation-model labs, not a foundation-model lab itself. Second, the Applied AI franchise explicitly covers public-sector and defence work, which is a different career proposition from the consumer-AI labs and a meaningful filter for international candidates. Third, the company's structure changed materially in mid-2025: Meta took a 49% non-voting stake at a roughly $29B valuation and founder Alexandr Wang moved to Meta to lead AI efforts, with Jason Droege named interim CEO (Scale blog, corroborated by Bloomberg and CNBC). For a candidate weighing Scale against the foundation-model labs, the live decision is whether the data / evaluations / public-sector platform fits the career thesis better than building the model itself, and whether the post-2025 leadership structure is one they want to join.