Notion
Model Behavior Engineer
About this role
Notion is hiring a Model Behavior Engineer to own the quality bar for AI products by designing evaluation systems, engineering prompts, and analyzing production data. You'll work across product and engineering teams to define quality standards, launch new models in partnership with leading AI labs, and shape how millions of users experience Notion's AI features.
What you'll do
- Design and iterate on system prompts and context strategies to improve AI product behavior
- Analyze production data, logs, and user feedback to identify and debug quality issues
- Build evaluation frameworks, datasets, and measurement systems to track AI quality over time
- Evaluate and launch new models from OpenAI, Anthropic, Google and other labs
- Communicate quality priorities and drive fixes across engineering and product teams
- Develop tooling and systems for AI observability and evaluation platforms
What they're looking for
- Driver mentality with bias to action
- Data analysis and SQL proficiency
- Experience with LLMs, prompting, or AI products
- Analytical thinking and problem-solving
- Clear written and verbal communication
- Curiosity about LLM capabilities and limitations
- Self-service data insights from large datasets
- Comfortable with ambiguity in fast-moving environments
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Notion
Notion builds an AI-powered collaborative workspace platform used by millions for productivity, featuring capabilities like search, automations, and AI-assisted features. The company is hiring software engineers for core product and AI development, technical support engineers for enterprise customers, data platform engineers for foundational systems, and security engineers for infrastructure and AI safety initiatives.
- Website
- notion.so
Likely interview questions
- Walk us through a time you debugged a complex problem in production data. How did you identify the root cause and what did you do about it?
- Describe your experience with LLMs and prompting. Have you iterated on prompts to improve model behavior, and what did you learn?