Why this role existsWe are scaling across client engagements: post-training dataset creation, benchmarking programs, and ongoing evaluation partnerships. Each project spans a different mix of AI labs, creative-tool companies, and research teams, with different requirements, timelines, and quality bars. We need someone to own this work from end to end.
Think of yourself as the operator for a project or vertical. You sit between three groups that do not naturally talk to each other: the client who needs the data, the creative experts producing it, and the internal engineering team building the tooling. Your job is to make all three feel like they are working on the same thing, and to ship results that are both rigorous and genuinely useful.
What you'll own- The methodology. Annotation protocols, rubrics, and evaluation designs for creative and multimodal outputs.
- Delivery. Scope, evaluator staffing, quality routing, timelines, and the client relationship, from kickoff to final readout.
- Quality. Data quality, inter-rater reliability, and real-time course correction when something feels off. You hold the bar, and you hold the levers to fix it.
- Analysis and storytelling. Qualitative and quantitative analysis turned into clear findings, recommendations, reports, and workshops that clients act on.
- The systems behind it. You turn repeated work into clean, measurable process, and you partner with engineering on the tooling that makes that process faster and more reliable.
- The narrative. Case studies and benchmark publications that sharpen how Contra Labs is understood in the market.
About you- You design studies for a living. You can scope a question, run the study, analyze the data, and translate it into a direction people can act on.
- You understand good creative, and you can explain it. You recognize creative quality and can say why, and you can tell the difference between personal preference and something closer to ground truth. You build frameworks that capture that signal faithfully.
- You care more about people than about technology for its own sake. You understand how humans judge creative work and design systems around that.
- You turn messy work into machines. You make process clear, predictable, and measurable, and you run a tight ship across several projects at once.
- High ownership, low ego, bias to action. You ship a v1, learn, and do what it takes to deliver.
Experience- 3+ years designing studies, running evaluations, analyzing data, and presenting findings.
- Experience with data labeling, annotation, human-in-the-loop systems, or AI and ML evaluation workflows.
- A track record managing several concurrent projects, including annotation, evaluation, or QA work.
- A real point of view on creative quality and what "good" looks like.
Bonus- A strong research background, qualitative and quantitative. A Master's or PhD in HCI, cognitive science, psychology, design research, or a related field is a plus, and so is equivalent depth from industry.
- Statistical methods for evaluation work: inter-rater reliability, sampling, regression, significance testing.
- RLHF, LLM evaluation, or benchmark design.
- Ethnography, UX research, or design-study methods.
- Startup or 0-to-1 experience.
Requirements- Based in San Francisco, California
Total comp- Salary: $150,000 to $200,000 USD plus equity
- Medical, dental, vision
- 401k matching
- Company laptop on your start date
Interview process- Recruiting team (20 minutes)
- CEO and co-founder (30 minutes)
- Technical interview, research and methods focus (45 minutes)
- VP of Product (45 minutes)
- Paid case study and presentation (60 minutes)
