Research Scientist

About Us

Our mission is to raise AGI with the richness of human intelligence — curious, witty, imaginative, and full of unexpected brilliance.

Surge was founded by engineers and researchers who dreamed of building the next generation AI. We're building a platform that powers the most powerful models in the world in partnership with companies like Anthropic, Google, Microsoft, and Meta.

At Surge, we believe the path to AGI isn't just about scaling compute—it's about embracing the unlimited ceiling of human intelligence and creativity in the data that shapes these systems. Our platform combines elite human expertise with cutting-edge tools for scalable oversight, from building rich RL environments to conducting rigorous evaluations that go beyond benchmarks. We've run a profitable business from day one without raising venture funding.

The Role

As a Research Scientist focused on data, you’ll work at the cutting edge of LLM development by designing the datasets that shape how frontier models behave. You’ll partner directly with AI research teams to experiment with new data collection strategies, evaluate dataset quality, and uncover insights that improve alignment, safety, and model performance.

This is a research-driven, impact-heavy role: you’ll have the opportunity to test ideas quickly in live production environments, shape data-centric methodologies for some of the world’s top AI labs, and help define how high-quality data fuels the next generation of AI systems.

What We’re Looking for

• Deep Curiosity About Data – Obsession with understanding how the structure, quality, and selection of data influence LLM performance

• Bias Toward Insight and Iteration – You think critically but move fast, designing lightweight experiments to generate actionable results

• Desire to Shape AI Development – Drive to build the foundational data layer that defines how the next generation of AI systems behave
  

What You'll Do

• Research data collection strategies and designing high-impact data slices that uncover model failure modes

• Model annotator behavior and designing experiments to optimize instruction clarity and reward signal reliability

• Develop metrics and frameworks for evaluating dataset quality, diversity, and impact on downstream model alignment