Software Engineer, Enterprise Agent Evals & Efficiency (AI)

Location: San Francisco, New York City, or Seattle, United States

Salary: $220K–$405K annually, plus equity. USD salary ranges apply only to U.S.-based positions; international salaries are set according to the local market.

Company

Perplexity develops AI-powered products and enterprise agents that automate complex work across connected tools, systems, and data.

What you will do

• Build realistic enterprise evaluation environments with connected applications, datasets, user roles, permissions, multi-step workflows, and reproducible system state.
• Own evaluation infrastructure, task suites, and automated graders calibrated against human judgment and end-to-end outcomes.
• Analyze agent trajectories and production failures, identify root causes, and ship improvements across the agent harness, tools, skills, and models.
• Develop reusable skills and improve tool descriptions, schemas, result handling, context management, caching, and execution strategies.
• Build model-selection and routing strategies that balance task quality, latency, reliability, and cost.
• Build regression gates and launch-readiness signals for model, connector, tool, skill, and harness changes, validating improvements in production.

Requirements

• 5+ years of software engineering or machine learning engineering experience, with ownership of complex production systems from design through deployment and operation.
• Strong system design skills, including reliable and scalable infrastructure for experimentation, evaluation, or data-intensive applications.
• Strong proficiency in Python, Go, or Rust and the ability to work across a multi-language production stack.
• Hands-on experience with LLM-based applications or agents, including tool use and multi-step workflows.
• Deep expertise in at least one area: evaluation infrastructure, benchmark design and automated grading; agent harnesses, tool interfaces and reusable skills; realistic multi-system test environments and simulations; or model selection, routing and inference efficiency including context management and caching.
• Strong experimental and statistical judgment, including representative task selection, meaningful success criteria, failure diagnosis, and distinguishing real improvements from noise.
• Sound technical judgment in balancing quality, latency, reliability, and cost, with experience turning evaluation findings into shipped improvements.
• Experience leading ambiguous cross-team initiatives from architecture through production adoption, setting technical direction, and improving engineering quality through mentorship.

Culture & Benefits

• U.S. full-time employees receive equity, health, dental, vision, retirement, fitness, commuter, and dependent-care benefits.
• Full-time employees outside the U.S. receive benefits tailored to their region of residence.
• Work focuses on measurable improvements to enterprise agent reliability, task success, latency, and cost.