Research Scientist Intern

What you'll do

• Identify data patterns that cause frontier models to perform poorly
• Create scoring systems and incentive structures for RLHF and RLVR model training
• Measure how dataset characteristics affect model performance and safety outcomes
• Collaborate with researchers to convert training goals into concrete data requirements

What they're looking for

• Genuine enthusiasm for understanding how data composition shapes model capabilities
• Comfort running rapid experiments and drawing conclusions from messy data
• Preference for hands-on work over abstract discussion
• Background in AI safety, model evaluation, or reinforcement learning environments is valuable

Pay and perks

• $10K - $17K monthly
• Company covers travel costs between offices
• Strong candidates may receive full-time employment offer