Location: San Francisco, CA (Onsite)
Compensation: $150,000 – $250,000 base | $250,000 – $450,000 total comp + equity
About AfterQuery
AfterQuery is an AI infrastructure company building training data and evaluation systems used by leading frontier AI labs. They work directly with top labs to improve model performance through high-signal datasets and experimentation. $30M raised at :$300M valuation. Founding team from Jane Street, Citadel, Google, Goldman Sachs, and Stanford AI Lab.
About The Role
This is a post-training research role focused on proving that data drives measurable improvements in model performance. You will design and run controlled experiments to isolate the impact of datasets on LLM behavior — working directly at the edge of model development. The focus is not theoretical research but building, running, and validating experiments that produce clear, defensible results tied to model capability improvements.
What You'll Own
- Design and run controlled SFT and RL experiments to measure dataset impact on model behavior
- Isolate performance improvements across reasoning, tool use, long-horizon tasks, and domain workflows
- Quantify lift and performance changes across capabilities; interpret messy experimental results
- Prove that specific datasets lead to measurable improvements under defined conditions
- Translate results into clear, defensible outputs for partner AI labs
- Collaborate with internal teams to iterate on dataset quality and build shared experimental infrastructure
- Communicate findings directly with partner AI labs; support relationship building and revenue through validated results
Requirements
- Strong familiarity with LLM training and evaluation methods including SFT and RL post-training
- Ability to design and execute lightweight experiments quickly
- Strong analytical instincts with ability to work through messy data
- Comfort working across domains including finance, software, policy, and enterprise workflows
- Bias toward building and execution over theory
- Undergraduate or master's research background preferred
Nice to Have
- Experience at RL environment companies, AI safety organizations, or benchmarking groups
- Experience running controlled training experiments end to end
- Published work in model evaluation, post-training, or data curation
- Strong software engineering skills alongside research experience
This Role Is NOT For
- Pure research profiles without hands-on execution or shipping experience
- Those who prefer narrow domain focus
- Candidates unable to operate in ambiguous, experimental environments
Logistics
- Role is fully onsite in San Francisco — please only apply if you can commit to this
- Multiple headcount with strong hiring demand
Shortlisted candidates will be contacted by David Joseph & Co., the recruiting partner managing this search on behalf of AfterQuery.