Learn the methodology of outcome-based evaluation that ships at Anthropic, Scale, Labelbox, OpenAI. Eight-week cohort. Spanish-English bilingual. Built and taught by a senior practitioner. Placement support included.
Apply nowThe most demanding agent-evaluation projects in the industry — OpenClaw Atlas, HiL-Bench, Blocker Injection, Model Response Evaluation — pay between USD 35 and USD 90 per hour for senior AI Trainers. The Spanish-speaking world has tens of thousands of qualified candidates and almost no structured training to reach those projects.
Orquor Academy was built because the founder ran these pipelines for the past two years and learned how the work actually flows. We are teaching it directly — the methodologies, the rubric design, the verifier discipline, the prompt engineering, the anti-detection writing — to the next generation of practitioners.
The market. The players. Who pays what for which kind of work. How outcome-based rubrics differ from old-style annotation. How to position yourself for the most demanding projects.
Atomic rubrics. Signed weights. Positive and negative verifiers. The 9-rule feedback decision tree. Real OpenClaw Atlas examples and worked exercises.
Writing pytest verifiers from scratch. Fixtures, parametrization, snapshot testing. How to structure a verifier suite that scales to 60+ tests across multiple domains.
Writing prompts and persona dialogue that does not get rejected by AI-detection. The 31-pattern checklist of AI-generated tells. Cultivating a human voice in technical text.
How to build a Universe (persona, world, tensions). Safety surface mapping. Private data, estate planning, financial advice, mental health — the domains where the work is most consequential.
The 12 weakness codes (INST / OVERENG / TOOL / LAZY / VERIFY / FALSE / ROOT / DESTRUCT / FILE / HALLUC / DOCS / VERBOSE). Scoring on the 0–7 scale. How to write strengths and weaknesses that survive QA.
Modifying Problem Statements, golden patches, test patches. The HiL-Bench framework. How to build tasks that surface real agent failures.
Direct introductions to the right teams at Outlier, Scale, Labelbox, Surge AI. Portfolio review. Interview preparation. Lifetime access to the Orquor Academy community.
Software engineers, data scientists, linguists, healthcare technical writers who want to enter the AI Trainer market.
You already work for Outlier or Scale at the basic-task level. You want to move into the senior projects that pay 2-3 times more per hour.
Physicians, lawyers, accountants, researchers in LATAM who want to apply their domain expertise to AI evaluation and earn at international rates.
Some technical comfort helps. You should be able to read Python at a basic level by the end of week 1. We do not require prior ML experience. We do require willingness to learn.
There is no guarantee. The placement tier provides curated introductions to specific projects, interview prep, and negotiation support. Whether a project hires you depends on your work. Our historical pass rate from Cohort applicants to first paid project is approximately 65% within 90 days of cohort completion.
Yes. Most cohort members do. Time commitment is 6–10 hours per week including the two live sessions.
The live sessions and Discord are bilingual — speak in either language. The recorded modules are in English with high-quality Spanish captions. Reading material is in English. The work you produce as practice will be evaluated in both languages.
Freddy [Apellido], founder of Orquor. He has been a senior AI Trainer for the past two years on OpenClaw Atlas, Labelbox World Slim, Outcome Ladybug, and Meter Pavilion. The curriculum is built from the methodology he uses daily.
Self-paced enrollment stays open. You also join the Cohort 02 priority list automatically. Cohort 02 opens 12 weeks after Cohort 01 completes.
14-day no-questions refund window on Self-paced. For Cohort live, full refund before week 2, 50% refund before week 4, no refund after week 4.