Research Scientist - Frontier Data

AfterQuery; San Francisco (CA); On-site; USD 250,000 - 450,000; Full time

Job summary

A leading AI research company in San Francisco is seeking a candidate to design datasets and evaluation frameworks that shape AI model training. Responsibilities include collaborating with research teams, running experiments, and developing metrics to assess model improvement. Ideal candidates are undergraduates or master's students with strong skills in data structure and evaluation methodologies. Total compensation ranges from $250k to $450k plus equity, reflecting the candidate's experience and contributions.

Qualifications

• Candidates should have research experience at the undergraduate or master's level.
• A strong interest in the impact of data quality on AI model behavior is crucial.
• Ability to work across diverse domains including finance, software engineering, and policy.

Responsibilities

• Design datasets and evaluation frameworks for AI models.
• Build and refine evaluation rubrics for training models.
• Run experiments to improve model capabilities.
• Develop frameworks for measuring dataset quality.
• Collaborate with research teams on training objectives.

Skills

Data structure manipulation

Experience in RL environment companies

Ability to design lightweight experiments

Strong quantitative skills

Education

Undergraduate or Master's degree in a relevant field

Job description

About AfterQuery

AfterQuery builds the training data and evaluation infrastructure that frontier AI labs use to make their models better. We work with the world's leading labs to design high signal datasets and run rigorous evaluations that go beyond static benchmarks. We are a small, early team (post Series A) where individual contributors have a direct impact on how the next generation of models learn and improve.

The Role

You’ll design the datasets and evaluation frameworks that shape how frontier models are trained and measured. Working directly with research teams at top AI labs, you’ll experiment with data collection strategies, diagnose model failure modes, and develop the metrics that determine whether a model is actually getting better. This is hands‑on, high‑leverage work: you’ll go from hypothesis to live experiment quickly, and your output will directly influence model training runs at scale.

What You’ll Do

• Design data slides and explore data shapes that expose meaningful model failure modes across domains like finance, code, and enterprise workflows
• Build and refine evaluation rubrics and reward signals for RLHF and RLVR training pipelines
• Model annotator behavior and run experiments to improve different model capabilities
• Develop quantitative frameworks for measuring dataset quality, diversity, and downstream impact on model alignment and capability
• Partner with lab research teams to translate their training objectives into concrete data and evaluation specifications

What We’re Looking For

• Great candidates are undergrad research or master’s research (but haven’t done a PhD)
• Major plus if they’ve worked for/interned for any RL environment companies in the past or any AI safety or benchmarking orgs like METR, Artificial Analysis, etc.
• Genuine obsession with how data structure, selection, and quality drive model behavior
• Ability to design lightweight experiments, move fast, and extract actionable insights from messy results
• Comfort working across domains (you’ll touch finance, software engineering, policy, and more)
• Strong quantitative instincts and familiarity with LLM training pipelines, RLHF/RLVR, or evaluation methodology
• A bias toward building over theorizing

Compensation Structure

$250k-450k total compensation + equity