Mercor — a profitable Series C AI data company valued at $10 billion, paying over $4 million per day to millions of domain experts who train frontier AI models — is accepting applications for its Research Fellowship from UAE-based candidates. This competitive programme funds researchers to build the next generation of AI benchmarks and evaluation techniques for Mercor’s APEX benchmark family — which measures whether frontier AI models can perform economically valuable professional work. Fellows receive a $40,000 (3-month) or $80,000 (6-month) stipend, unlimited API credits, dedicated GPU compute budget, expert human data time, weekly mentorship, and access to frontier model APIs and real enterprise evaluation problems from Fortune 500 and frontier lab partners.
About Mercor — Organising Human Intelligence to Power the AI Economy
Company: Mercor — a profitable Series C company valued at $10 billion — the leading layer between human expertise and frontier AI models
Scale: Millions of domain experts paid over $4 million per day to train frontier AI models — the world’s largest human intelligence network for AI data
APEX Benchmark Family: Measures whether frontier AI models can do economically valuable professional work — investment banking, corporate law, accounting, software engineering, and graduate-level science
Enterprise & Labs: Mercor Enterprise serves Fortune 500 companies; APEX results are used by frontier labs to measure and improve their most capable AI models
Why This Mercor Research Fellowship Stands Out in UAE 2026
$80K Stipend + Resources: One of the most generously resourced AI research fellowships available globally — stipend, GPU compute, unlimited API credits, and paid human expert data
Real Industry Impact: Your benchmark shapes how the AI industry measures frontier model capability — published as a paper, open dataset, APEX leaderboard, or adopted methodology
Frontier Lab Access: Work directly with the APEX research team, access real enterprise evaluation problems from Fortune 500 and frontier-lab partners, and build connections across the AI research ecosystem
Full-Time Job Pathway: Standout fellows are considered for a full-time offer on the APEX research team — making this fellowship the highest-value AI research career entry point in 2026
Fellowship Overview
The Mercor Research Fellowship funds selected researchers to build the next generation of AI benchmarks and evaluation techniques for the APEX benchmark family — which measures frontier AI model performance on economically valuable professional work. Fellows pitch a specific benchmark or evaluation technique they want to build, and if selected, receive the time, compute, expert labour, and mentorship to design, implement, and release it end to end. Fellows work directly with the APEX research team, get access to real enterprise evaluation problems from Mercor’s Fortune 500 and frontier-lab partners, propose and scope new benchmark domains, design task specifications and grading rubrics with vetted domain experts, build and validate benchmarks through piloting and stress-testing, run frontier models against their benchmark, and publish results as a paper, open dataset, APEX leaderboard entry, or adopted methodology.
Why This Fellowship Matters: As a Mercor Research Fellow from the UAE, you contribute to one of the most consequential unsolved problems in artificial intelligence — how do we actually know whether a frontier AI model can perform economically valuable professional work reliably, at scale, without shortcuts or contamination? Every APEX benchmark you design and validate directly influences how GPT, Claude, Gemini, and the next generation of frontier models are measured, compared, and improved. At a company paying $4 million per day to domain experts and serving Fortune 500 companies with enterprise AI evaluation infrastructure, your research is not a thesis exercise — it is the measurement layer that the entire AI industry depends on.
What You Will Do as a Mercor Research Fellow
Benchmark Proposal, Scoping & Design
- Propose and scope a new benchmark or evaluation technique in a domain APEX doesn’t yet cover well — or design a meaningfully harder, more robust version of an existing APEX domain benchmark
- Design task specifications and grading rubrics in partnership with Mercor’s network of vetted domain experts — lawyers, accountants, engineers, scientists, and consultants who validate real-world task difficulty and answer quality
- Define clear success criteria, evaluation metrics, and grading methodology that captures the genuine economic value of the professional work being assessed
Benchmark Build, Validation & Frontier Model Testing
- Build and validate the benchmark through systematic pilot testing — calibrating scoring rubrics, stress-testing for data contamination, and identifying gameable shortcuts that would invalidate model comparisons
- Run frontier AI models against your benchmark — systematically analysing where and why state-of-the-art models fail to perform economically valuable professional work reliably
- Implement contamination-resistance techniques and evaluation methodology improvements — contributing to the field’s understanding of how to design benchmarks that remain valid as models improve
Research Publication & APEX Integration
- Publish your benchmark and findings — as a research paper, open dataset, a new leaderboard on APEX, or a methodology the APEX team adopts internally for ongoing frontier model evaluation
- Partner with Mercor’s research and engineering teams to fold your insights back into the APEX public benchmark family — ensuring your fellowship outputs have lasting impact beyond the fellowship period
- Present your benchmark design rationale, methodology, and results to the APEX research team and Mercor’s broader research organisation — building research communication skills alongside technical evaluation expertise
What Mercor Looks For in Fellows
Essential Profile
- Genuine interest in evaluation as a research discipline — not as a stepping stone to a model-building role, but as a primary intellectual and professional focus in its own right
- A specific, well-scoped idea for a benchmark or evaluation technique you want to build — the fellowship is funded around your pitch, not a generic research rotation or exploratory proposal
- Background in Computer Science, ML, statistics, or an adjacent field such as measurement, psychometrics, HCI, or social science — no requirement to have published in ML venues
- Comfortable in a startup environment — fast iteration, direct access to real customer problems, less hand-holding than an academic research lab
- Able to commit at least 20 hours per week for the duration of the fellowship — during a leave of absence, over a summer period, or a flexible stretch of a PhD
Strong Advantages
- Experience with agentic evaluation, RL environments, or evaluation of multi-step AI agent systems
- Domain expertise in law, finance, medicine, accounting, or a scientific field — enabling authentic, expert-validated benchmark design
- Prior experience designing or critiquing AI benchmarks, evaluation datasets, or assessment methodology in any professional or academic context
- Published research or open-source contributions in AI evaluation, NLP benchmarking, or related areas
About Mercor — The Layer Between Human Expertise and Frontier AI
Mercor’s mission is to organise human intelligence to power the AI economy. As a profitable Series C company valued at $10 billion, Mercor has built the leading layer between domain expert knowledge and the frontier AI models that millions of people and organisations depend on. With millions of experts on the platform earning over $4 million per day for training frontier models, Mercor’s APEX benchmark family measuring AI’s real-world professional impact, and Mercor Enterprise bringing this infrastructure to Fortune 500 companies — the Mercor Research Fellowship offers UAE-based researchers a genuinely rare opportunity to contribute to one of the most important unsolved problems in artificial intelligence: measuring, honestly and rigorously, whether frontier AI can actually do the work the world needs it to do.
Career Excellence: Design the AI benchmarks that shape how GPT, Claude, and frontier models are measured — $40K–$80K Mercor Research Fellowship, remote from UAE.
Who Should Apply?
- PhD Students & Graduate Researchers — UAE: With a specific AI benchmark or evaluation methodology idea — able to commit 20+ hours per week during a PhD stretch, summer, or leave of absence
- AI Evaluation & Benchmark Specialists: With experience designing, validating, or critiquing AI benchmarks — genuinely interested in evaluation methodology as a primary research focus
- Domain Experts with AI Literacy: Lawyers, accountants, engineers, financial analysts, or scientists with strong AI/ML literacy and a compelling idea for measuring AI capability in their professional domain
- ML Researchers — Statistics & Measurement Backgrounds: With psychometrics, HCI, social science, or statistical measurement expertise — bringing rigorous evaluation methodology from adjacent disciplines into AI benchmark design
- UAE-Based AI Researchers & Practitioners: Ready for a remote-first funded research fellowship that combines real-world impact, frontier lab access, and a clear pathway to a full-time AI research role at a $10B company
Recently Opening Job👇
Manager Digital Products Marine Services Jobs Abu Dhabi UAE 2026