Member of Data Staff (Data Acceleration)

Perplexity AI is a company built around the belief that artificial intelligence should meet the highest standards of performance and reliability. The organization is at the forefront of AI development, focused on creating tools and systems that help people accomplish more. Perplexity's data team is pioneering an AI-native approach to analytics and data science work, and you would join this effort as a key contributor shaping how modern data organizations operate at scale.

Data Staff Member - Data Acceleration

Perplexity seeks an experienced data professional to build AI systems that transform how the data team works. You'll design and implement agents and internal tools that automate end-to-end analysis workflows, from hypothesis formation and query execution through result interpretation and recommendation generation. This role goes beyond query assistance to create reliable, production-grade systems that handle complex data science tasks with appropriate human oversight and evaluation frameworks.

What you'll do

• Develop AI agents capable of independent data exploration, hypothesis testing, query execution, result analysis, and actionable recommendation generation with built-in review mechanisms.

• Construct retrieval infrastructure and evaluation loops that enable AI systems to query data warehouses accurately using semantic context and metadata.

• Transform validated AI workflows into standardized, repeatable systems and shared tools that the entire data team can adopt and leverage.

• Engineer automated data quality systems including self-healing pipelines, generated and validated dbt models, and diagnostic workflows that reduce manual intervention.

• Create AI-powered agents that analyze experiment results, identify statistical issues, surface likely drivers, and produce ship or no-ship assessments.

• Build internal data products that replace repetitive ad hoc requests with self-serve AI interfaces stakeholders use daily.

• Take ownership of the complete development cycle from identifying high-impact problems through prototyping, accuracy evaluation, UX design, production deployment, and ongoing quality monitoring.

What you'll bring

• Six or more years of hands-on experience in data science, analytics engineering, data engineering, or closely related work, with deep familiarity with real-world data challenges.

• Advanced SQL proficiency and strong analytical judgment to evaluate metrics, experiments, data models, and warehouse architecture without relying on automated tools.

• Proven product instinct that translates technical capabilities into adoptable workflows and systems people genuinely use.

• Practical experience building with large language models, agents, retrieval-augmented generation systems, evaluation frameworks, or AI-assisted workflows.

• Solid foundation in data pipeline architecture, dbt, warehouse schema design, data quality approaches, and tradeoffs in building maintainable data systems.

• Systematic mindset that turns manual processes into scalable solutions; you ship iteratively, measure outcomes, and refine based on results.

• Comfort with significant autonomy in a newly formed function where you'll help chart the roadmap alongside execution.

Nice to have

• Prior experience developing production AI agents or evaluation systems for agents.

• Background with Snowflake, semantic layer platforms, or metadata management systems.

• Track record building and shipping internal tools, Slack integrations, command-line interfaces, or developer productivity software with real adoption.

• Strong experimental design background including metric definition and statistical validation.

• Experience with business intelligence tools and judgment about what warrants automation versus human review.

What they offer

• Full-time position based in San Francisco.

• Work at an AI-focused company with access to frontier models and infrastructure from day one.

• Small, empowered team structure enabling rapid iteration from concept to production with minimal approval layers.