Olympiad-Level Problem Setter - Physics / Chemistry / Biology / Maths (AI Training Data)

We are Scaler AI Labs, and we are building a high-difficulty question bank to train and evaluate next-generation AGI models. Frontier AI systems have already reached gold-medal performance on standard olympiad problems. To push further, we need problems that are genuinely beyond what current AI can solve - problems that require real creative insight, not pattern matching or technique application. That's exactly what this project is about. What the work involves: - Creating original problems with complete model solutions and marking schemes - Reviewing existing problems for correctness, originality, and difficulty - Flagging duplicates or anything solvable by standard techniques or AI Subjects we need: - Mathematics - IMO P3/P6 equivalent. Non-routine, requires a non-obvious key insight - Physics - IPhO T3 equivalent. Multi-domain theoretical, not formula-based - Chemistry - IChO hardest theoretical. Multi-step mechanisms, synthesis, physical chemistry - Biology - IBO hardest theoretical. Cross-domain reasoning, novel data interpretation Who we are looking for: - Past IMO / IPhO / IChO / IBO contestants — gold or silver level preferred - National or international olympiad coaches - PhD / postdoc researchers with a strong competition background - Problem setters for national or international olympiad committees - If you have already worked on AI training data projects involving hard science or math problems - you already know what we need, and we'd love to have you on board Experience with this kind of work is a big plus. But if you have the olympiad depth and haven't done AI training data before, that's completely fine too. Engagement: - Per-problem payment, negotiated based on subject and complexity - Fully remote, flexible hours - All work under NDA To apply, please mention: - Your competition or coaching background - medals, roles, committees - Which subject(s) you are applying for - One sample problem you have authored or a pointer to published work - Roughly how many problems you can contribute per day/week - If you have worked on AI training data projects before, briefly describe that experience Your work here directly contributes to where AGI development goes next. If this sounds like your kind of challenge, we'd love to hear from you.