Applied AI Engineer — Frontier Agent Capabilities

Perplexity; San Francisco (CA); On-site; USD 220,000 - 405,000; Full time

Job summary

Perplexity Computer is seeking a senior software engineer to join the Agent Capabilities team. You will evaluate frontier AI breakthroughs, design scalable agent capabilities, and own the lifecycle from prototyping to production.

You will work across PM, Data Science, and Research to transform model advances into reliable user experiences. You will shape architecture and product experiences, ensure safety and performance, and drive measurable impact across complex real-world tasks with a

Qualifications

• 6+ years of professional software engineering experience.
• Experience owning the AI product lifecycle, including data analysis, evaluation, monitoring, and iterative improvement.
• Practical experience in agent harnesses, tool use, context engineering, model evaluation, or browser automation.

Responsibilities

• Evaluate frontier models against real user tasks and turn advances into production agent systems.
• Improve agents’ ability to plan, use tools, manage context, and complete long-running tasks reliably.
• Apply ML/LLM techniques to design scalable agent capabilities and production-ready experiences.
• Own agent behavior end-to-end from user interfaces to backend services; define evaluations for success.
• Build secure, observable agent systems with tracing, monitoring, and safeguards.

Skills

Frontier AI interest

Ownership mindset

Product judgment

Ambiguity-to-solution

Communication skills

Tools

Python

Go

Rust

PostgreSQL

DynamoDB

AWS

TypeScript

Job description

Perplexity Computer is seeking a senior software engineer to join the Agent Capabilities team. You will evaluate frontier AI breakthroughs, design scalable agent capabilities, and own the lifecycle from prototyping to production.

You will work across PM, Data Science, and Research to transform model advances into reliable user experiences. You will shape architecture and product experiences, ensure safety and performance, and drive measurable impact across complex real-world tasks with a