Where to actually get AI evaluation work. Pay and fit notes are hand-curated and honest. Link and signup status refresh weekly. This is a read-only snapshot.
| Platform | Signup | Pay notes | Why it fits | |
|---|---|---|---|---|
| ⚪ | DataAnnotation.tech | not checked | Reported as one of the better crowd-AI payers, but rate varies by project. Verify on offer. | LLM training, coding, and reasoning tasks. Coding background helps. Entry is a paid trial. |
| ⚪ | Mercor | not checked | Project and seniority dependent. Verify on offer. | Matches experts to AI-lab projects via interview. A strong GitHub profile is a real advantage here. |
| ⚪ | Mindrift (Toloka) | not checked | Per-project rates. Comes in waves. Verify on offer. | Freelance AI-tutor projects for domain experts including coding and analytics. Flexible. |
| ⚪ | Surge AI | not checked | Reputable payer. Selective intake. Verify on offer. | High-quality RLHF and evaluation data. Selective, good reputation, rubric-style work. |
| ⚪ | Turing | not checked | Varies by role and region. Verify on offer. | AGI/LLM training plus coding expert work. Dev-leaning, so your repos are an asset. |
| ⚪ | Kaggle | not checked | Prize-based, not wages. Some comps have large cash pools; many are for medals and ranking. | ML and data-science competitions. Strong fit for your analytics background. Open to anyone. |
| ⚪ | DrivenData | not checked | Prize-based. Cash prizes on many competitions. | Data-science-for-good competitions. Clear problems, open entry, good fit. |
| ⚪ | Devpost | not checked | Prize-based, often cash plus cloud credits. Varies per hackathon. | Hackathon hub. AI-agent and DevOps themed events run constantly; your gatekeeper and Compass work fits well. |
| ⚪ | RWS / Article One Partners | not checked | Per-study rewards / bounties for accepted prior art. Not hourly. | Patent prior-art research (Article One Partners, now RWS). You actively submit studies here. This is NOT the RWS/Appen crowd AI-eval that was excluded. |
| Platform | Signup | Pay notes | Why it fits | |
|---|---|---|---|---|
| ⚪ | Toloka AI | not checked | Per-task. Expert tracks pay more than legacy microtasks. Verify on offer. | Older crowdsourcing platform that pivoted into LLM eval and expert tasks. |
| ⚪ | Invisible Technologies | not checked | Decent by reputation. Role dependent. Verify on offer. | AI training and LLM QA work for labs. More ops-flavored. |
| ⚪ | Defined.ai | not checked | Project dependent. Verify on offer. | Speech, text, and LLM data marketplace. Corporate clients, slower onboarding. |
| ⚪ | iMerit (Ango Hub) | not checked | Enterprise contracts. Verify on offer. | Annotation plus expert tasks. More structured and enterprise. |
| ⚪ | Innodata | not checked | Enterprise. Verify on offer. | Enterprise AI data prep with expert annotator programs. |
| ⚪ | Sapien AI | not checked | Newer, rates unproven. Verify on offer. | Crowd labeling and RLHF, newer player. |
| ⚪ | Pareto.ai | not checked | Startup, rates vary. Verify on offer. | Data-labeling startup doing expert and RLHF tasks. |
| ⚪ | Micro1 | not checked | Placement dependent. Verify on offer. | AI-driven vetting and recruiting that places experts into AI data work. |
| ⚪ | AIcrowd | not checked | Prize-based. Varies by challenge. | AI and ML research challenges. More research-flavored. |
| ⚪ | Zindi | not checked | Prize-based. Cash prizes on many. | AI competitions, Africa-focused but open globally. Real-world datasets. |
| ⚪ | HackerEarth | not checked | Prize-based; some challenges are hiring-linked. | Hackathons and hiring challenges. Mixed dev and ML. |
| ⚪ | Topcoder | not checked | Pays per challenge. Competitive fields. | Algorithm, development, and design challenges that pay out. More traditional dev. |
| Platform | Signup | Pay notes | Why it fits | |
|---|---|---|---|---|
| ⚪ | Clickworker / UHRS | not checked | Low, per-task. Filler work. | Microtask crowd, some AI tasks. Filler-tier pay but always-on. |
| ⚪ | Anthropic | not checked | Strong when a role exists, but selective and intermittent. | A company careers page, not an open application. Safety/red-team/contractor data roles show up only now and then. Watch, do not expect a door that is always open. |
| ⚪ | OpenAI | not checked | High when active, very selective. The old /contributions link is unreliable; use careers. | Also a company careers page, not an open application. Runs expert AI-training and contractor programs now and then; search "expert" or "AI trainer". Very selective. |
| ⚪ | Prolific | not checked | Per-study rates set by researchers. Easy entry, uneven volume. | Research-study platform. Hosts AI/LLM behavior experiments beyond surveys. Easy to start. |
| ⚪ | Humanloop | not checked | Not gig work. Salaried roles only. | NOTE - this is an LLM eval/dev tooling company, not a gig platform. Watch for full-time/contract roles. |
| ⚪ | Major League Hacking (MLH) | not checked | Mostly prizes and swag, not cash-heavy. | Open hackathon season. Student-leaning, but anyone can join member events. |
| ⚪ | ARC Prize | not checked | Prize-based, not wages. Big pool, very hard. | Reasoning benchmark with a large prize pool. Hard, but squarely in your wheelhouse. |