Member of Technical Staff, Post-Training Research & Data

Intelligence•San Francisco, CA

9d•Onsite

About The Position

Intelligence Mission is a product lab company focused on advancing human and superintelligence. They are the creators of DesignArena, a platform with over 5.2 million users in 8 months, recognized by prominent figures in AI. The company boasts a highly talented team, significant venture backing, and is considered one of the fastest-growing seed-stage startups in San Francisco. Intelligence Mission works with leading AI model providers like OpenAI to evaluate state-of-the-art multimodal models across various domains. Their flagship product, Design Arena, is a widely referenced benchmark for AI-generated visuals, and Prediction Arena was a pioneering platform for autonomous model trading. Social Arena tested AI models' ability to manage social media engagement.

Requirements

Data-pilled. Understand that data quality is the real bottleneck to superintelligence.
Strong STEM background. Studied Computer Science, Machine Learning, Data Science, Statistics, Math, Engineering, Physics, or a related field.
Experience building ML systems. Have trained preference models, built data pipelines, or worked on ML infrastructure that runs in production.
Interested in how AI learns from human feedback. Excited by solving problems at the intersection of human-model interaction.

Nice To Haves

Opportunity to be forward-deployed and work directly with researchers at frontier labs to creatively scale new model capability strategies for improvement.

Responsibilities

Invent new ways to detect and scale the data that makes frontier AI models smarter.
Develop scalable systems that generate, curate, and improve high-quality supervision across coding, multimodal, and agentic tasks.
Train and improve preference, reward, and ranking models from millions of human interactions.
Develop infrastructure for large-scale experimentation, model training, and specialize in online evaluation techniques.
Design systems that transform human preference data into reliable signals for downstream model evaluation and training.