Member of Technical Staff (Software Engineer, Capabilities)

Perplexity • San Francisco, CA

3d • $220,000 - $405,000

About The Position

In 2026, we launched Computer, the Perplexity agent that turns knowledge into action. The Capabilities team builds the Skills, Workflows, and Artifacts that users and agents reuse to do better work. As a Member of Technical Staff on the Capabilities team, you'll own systems spanning the agent product, core engine and the evaluation of Skills and Artifacts. Our goal is for Computer to be the natural place to use each new model capability as it reaches the frontier, and you'll help define how that happens.

Requirements

• 8+ years of professional software engineering experience, shipping and owning complex systems end-to-end.
• Strong backend / Full Stack engineering skills, with experience in designing and building scalable and reliable distributed systems, serving high traffic and a large user base.
• Demonstrated technical leadership: you scope ambiguous problems, set direction, and drive cross-team projects to durable outcomes.
• Strong product judgment and ownership instincts; you turn vague needs into simple, reliable systems and ship without waiting for perfect specs.
• Comfort with data-informed decisions; you define the metrics and evals that prove a system works, and iterate on them.
• Genuine interest in AI products, with hands-on adoption and a willingness to learn quickly.

Nice To Haves

• Experience building agentic systems (tool calling, subagents, long-running or autonomous task execution).
• Experience building developer platforms or reusable-capability primitives (SDKs, plugin systems, workflow engines).
• Experience with evaluation, benchmarking, or quality systems for ML/LLM-powered products.
• Time spent at a fast-growing startup or on a high-ownership engineering team.

Responsibilities

• Build a deep, hands-on understanding of how frontier LLMs reason and where they break, then turn that into better engine and primitive design.
• Bring each new model capability into the product as it reaches the frontier, including planning, long-running tasks, autonomous execution, and self-evolving skills and agents.
• Work closely with the agent engine layer (context management, tool calling, planning, long-horizon execution) to turn frontier capabilities into reliable primitives.
• Design, build, and own the primitives at the heart of Computer (Skills, Workflows, and Artifacts) so they compose into one coherent experience.
• Build the evaluation systems (benchmarks, evals, rubrics, feedback loops) that make each capability best-in-class before broad rollout.
• Set technical direction on ambiguous problems and raise the bar through design reviews, mentorship, and example.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior
• Education Level: No Education Listed