Scale AI has built one of the widest entry-to-career bands in the US AI labor market — from $15-an-hour generalist annotation tasks to $220,000 total compensation for a new-grad software engineer. The path between those two numbers is not automatic, not codified, and not a formal conversion program. But for a specific cohort of graduates — those with CS, math, or coding credentials who entered through Scale's Outlier contractor platform — it is becoming a documented route into full-time AI employment. Understanding exactly how that route works, where it breaks down, and what it pays at each stage is the real story behind Scale's position in the 2026 graduate hiring market.
The picture that emerges from public filings, compensation data, job postings, and employer statements is more complicated than the narrative of a seamless pipeline. It is also more structurally significant.
What the Funnel Actually Looks Like
Scale AI's workforce operates in two distinct tiers that run on separate tracks.
The first tier is the Outlier contractor network — the platform previously branded partly as Remotasks — which operates as a global task marketplace for AI training data. As of 2026, Scale's total workforce across both full-time employees and contractors sits between 4,000 and 6,000, with full-time employees representing approximately 1,000 to 1,200 of that count. The contractor base, which handles annotation, RLHF data generation, and model evaluation work, runs into the hundreds of thousands globally. The company itself has described its Outlier network as comprising contributors who often hold advanced degrees.
Compensation within Outlier spans a wide range by task type. Generalist content evaluation tasks pay approximately $15 to $22 per hour, with basic labeling roles at the lower end of that band. Coding RLHF work — writing and evaluating code prompts for large language model training — pays $50 to $65 per hour for computer science specialists. PhD-level domain expert work in fields like medicine, law, and mathematical reasoning commands rates well above $100 per hour on specialized projects. The data labeling market overall hit $1.69 billion in 2025, growing 154 percent year over year, and Scale commands the largest single share of that market.
The second tier is Scale's full-time hiring operation, which is lean, selective, and structured around two entry points for new graduates: the Strategic Projects Lead program and the Software Engineer new grad track.
The SPL program — currently recruiting for its 2026 cohort — is Scale's most accessible formal full-time entry point. Base salary in Scale's hub locations of San Francisco, New York, and Seattle runs $112,000 to $140,000, with equity on top in the form of stock options granted at board approval. The role draws from operations, data science, and STEM backgrounds, and explicitly rewards adaptability over credentialed experience. The SWE new grad track targets candidates with CS, EECS, or statistics degrees and prior engineering internship experience. Total compensation for new-grad software engineers at Scale, based on Levels.fyi data, runs approximately $175,000 to $220,000.
The Gap Between the Tiers — and How Some Graduates Cross It
Here is where the conversion pipeline story becomes specific, and also contested.
There is no documented formal program at Scale that converts Outlier contributors into full-time employees. Scale does not publish conversion statistics. There is no public statement from Scale AI leadership — from Alexandr Wang, who left the CEO role in June 2025 to join Meta, or from current CEO Jason Droege — committing to a structured contractor-to-staff track.
What does exist is a documented career signal: hiring managers at AI labs, including but not limited to Scale itself, have begun treating sustained, high-volume Outlier work at the coding or domain expert tier as a meaningful resume indicator. For a new grad who did not land a direct full-time role at a frontier lab but spent six to twelve months generating coding RLHF data at the $50 to $65 rate — producing annotated outputs at sufficient quality and volume — that work establishes demonstrable, production-proximate AI experience. It shows up on a resume as time spent building the training data behind models in active production.
The comp trajectory from that starting point to full-time employment is meaningful in dollar terms. A computer science graduate working consistent coding hours on Outlier at $50 per hour, 25 to 30 hours per week, generates $65,000 to $78,000 annualized before taxes. That is not career compensation. It is, however, more than the median entry-level data labeler salary of approximately $60,400 per year in the US, and it is a bridge position with a legitimate pathway to the next stage — applying for Scale's SPL program or SWE new-grad track, or using the work as a differentiator when applying to frontier lab roles at Anthropic, OpenAI, or Cohere.
The Classification Fault Line
Scale's workforce structure came under formal legal pressure in 2025 in a way that directly bears on how graduates should read this pipeline.
In December 2024 and January 2025, two former Scale AI workers filed separate wage lawsuits alleging they were misclassified as independent contractors rather than employees, denying them access to overtime pay, sick leave, and other statutory protections. The January 2025 complaint named former contributor Amber Rogowicz and alleged that Scale's effective pay rate worked out to approximately $15 an hour, in alleged violation of California minimum wage standards.
The US Department of Labor opened a formal compliance investigation in March 2025, serving Scale AI with a subpoena seeking overtime payment records. The investigation also targeted Scale's HR staffing partners HireArt and Upwork, indicating scrutiny of the full engagement chain.
The DOL dropped the investigation in May 2025 without enforcement action. The timing coincided with a broader shift in Labor Department enforcement posture on contractor classification rules under the incoming administration. Scale confirmed it was "pleased with the decision."
For graduates evaluating the Outlier on-ramp, the legal episode has a practical implication: the work is contractor work, not employee work, and the distinction has direct consequences for benefits access, tax treatment, and the absence of any formal conversion obligation. Scale is not required to convert strong-performing Outlier contributors into full-time employees. It sometimes does; there is anecdotal evidence of contributors who moved into full-time roles after demonstrating sustained quality work. But the pathway is discretionary, not programmatic.
What Jason Droege Said About Hiring
CEO Jason Droege, speaking on Lenny Rachitsky's podcast in October 2025, outlined his three-trait hiring framework for Scale's full-time staff: curious problem solvers, humble collaborators, and effective leaders. On the question of experience versus adaptability, Droege was direct: "The world's changing, right? So you do need people that are adaptable. So all the experience is not necessarily one-to-one relevant." He made one explicit exception — for research roles, he said, specific domain experience "is critical because the market's moving so fast, you don't have time to train up some people."
That framing maps onto Scale's actual new-grad architecture. The SPL program is structured for adaptable generalists — people who can navigate ambiguity inside a fast-moving AI infrastructure company. The SWE and research tracks demand demonstrated technical depth. Droege did not, in public statements, describe a formalized contractor conversion track. What he described is a company that values learning velocity over credential accumulation — a cultural signal that is relevant to how Scale evaluates candidates who have spent time in the Outlier network.
How Scale Compares to Big Tech Campus Recruiting
The contrast with traditional Big Tech graduate hiring is significant on two dimensions: structure and volume.
Google, Meta, Microsoft, and Amazon operate large-scale campus recruiting operations with structured internship-to-full-time conversion programs, defined offer timelines, fixed signing bonuses, and published base salary bands. Meta's L3 new-grad engineering offer, for instance, runs approximately $177,000 to $214,000 in total compensation based on Levels.fyi data, with a structured four-year equity vest and a first-year cliff. Google's L3 new-grad package is comparable. These programs generate thousands of full-time offers each year across hundreds of university partner programs.
Scale's new-grad operation is narrower by every measure. The company has approximately 1,200 full-time employees total — smaller than many individual engineering teams at the Magnificent Seven. Its formal new-grad hiring is measured in dozens per cohort, not thousands. The selectivity is higher by ratio, but the absolute pipeline is small.
Where Scale differs structurally is in the existence of the Outlier entry point at all. Big Tech does not offer a contractor-facing marketplace that serves as an informal on-ramp to full-time employment. Amazon's Mechanical Turk has existed for two decades and has never functioned as a hiring pipeline. Scale's setup is different because Outlier tasks, particularly at the coding tier, generate the kind of work product — documented AI training contributions in Python and related languages, model evaluation outputs, RLHF preference rankings — that carries direct signal for a hiring manager evaluating a new-grad engineering candidate.
That asymmetry is real, and it is part of why Scale's hiring profile for 2026 is worth tracking separately from the standard Big Tech new-grad stack ranking.
What's Next
Three things to watch in the second half of 2026:
The DoD expansion hiring signal. Scale holds a production agreement with the Pentagon's Chief Digital and AI Office with a $500 million ceiling, confirmed by Bloomberg in May 2026. The Washington D.C. office expansion and the company's stated plan to hire 200 people across enterprise, public sector, and international public sector roles means the fastest-growing entry point at Scale in H2 2026 is forward-deployed engineering and government account roles — not the Outlier-to-full-time path. New grads with security clearance eligibility or defense-relevant technical backgrounds are the target cohort.
The wage classification floor. The DOL investigation was dropped, but the two private wage lawsuits remain in the court system. If either reaches settlement or judgment, Scale may face structural changes to how it classifies and compensates its highest-volume Outlier contributors — changes that could either formalize the conversion pipeline or harden the wall between contractor and employee status.
Scale Labs as a research entry point. Scale's March 2026 launch of Scale Labs — its expanded research division covering AI evaluation, frontier safety, and post-training capabilities — creates a new full-time entry point for research-oriented graduates that sits outside both the SPL track and the traditional SWE path. Scale Labs is publishing, not just shipping; its research posture is closer to a frontier lab than to the infrastructure company Scale was in 2022. For a PhD or strong master's graduate in ML, alignment, or evaluation research, Scale Labs is a distinct option in the 2026 market that did not exist 18 months ago.
Scale AI enters the back half of 2026 as a company running two parallel workforce systems that do not formally connect — and the gap between those systems is the most productive place for new graduates to apply pressure. The Outlier network pays real money at the coding tier and generates legitimate AI work experience. The formal new-grad programs are selective but well-compensated. The conversion from one to the other is not guaranteed, not published, and not easy. For graduates willing to treat contractor work as a deliberate investment rather than a fallback, Scale is the only company in the US AI market where that investment can plausibly convert.