About the Role
The AI Acceleration team’s mission is to make Perplexity the best place to build and ship software with AI agents. We build the infrastructure that engineers and agents rely on, and develop new ways for agents to write, test, review, and ship code. Our work helps teams move from an idea to production faster, without compromising quality or reliability.
We own the end-to-end development experience at Perplexity, from the environment where code is written to the systems that ship it to production. That includes local and cloud coding environments, build and test infrastructure, code review, and release management. We’re also building autonomous software development systems, where agents take on increasingly complex engineering work with explicit evaluation, verification, and safety controls. We own both the infrastructure these systems depend on and the workflows that make them useful.
As an AI Acceleration Engineer, you’ll turn emerging agent capabilities into dependable systems used across Perplexity. You’ll work directly with engineers to understand where development slows down, experiment with new approaches, and take promising ideas through to production. This role combines systems engineering, developer experience, and hands-on agent development, with a direct measure of success: helping teams spend less time on delivery overhead and more time building products.
Responsibilities
Identify, prioritize, and deliver the highest-impact opportunities to accelerate how Perplexity builds and ships software with AI.
Work directly with engineering and business teams to understand their workflows, constraints, and bottlenecks, and build solutions that address how they actually work.
Build and operate the infrastructure that engineers and agents depend on, including development environments, build and test systems, code review, and release tooling.
Make our codebases, systems, and internal knowledge easier for agents to navigate, understand, and use through better tooling, interfaces, documentation, and context.
Take agent-driven workflows from prototype to production, building the evaluations, verification, and controls needed to delegate increasingly complex engineering work safely.
Improve delivery speed and software quality together. Measure the impact of your work on development time, review effort, and production reliability, and use those results to guide what comes next.
Evaluate emerging models, tools, and approaches through hands-on experimentation. Turn the evidence into clear technical decisions, and drive adoption of the technologies that prove useful.
Use AI extensively in your own engineering work, share what works, and help colleagues adopt effective practices. Partner with engineering leadership and recruiting to strengthen how we hire and develop engineers who work effectively with AI.
Qualifications
5+ years of software engineering experience, ideally with substantial work in infrastructure, platform engineering, or developer productivity in technically complex environments.
Daily, hands-on use of AI models and coding tools, with practical judgment about where they perform well, where they fail, and how to verify their output.
Strong proficiency in one of the languages – Python, Go or Rust, along with the willingness to learn other languages as the work requires.
Broad familiarity with the systems used by AI product and applied research teams, including cloud infrastructure, distributed systems, data pipelines, and ML tooling. You don’t need deep expertise in every layer, but you should understand how they fit together.
An ability to understand how engineers and researchers across disciplines work, paired with curiosity about the needs of business and operations teams. You build tools around users’ actual workflows and constraints.
A drive to make codebases, systems, and processes easier to understand, operate, and change. You look for ways to replace brittle workflows and implicit knowledge with clear interfaces and reliable automation.
Sound technical judgment about when to experiment, when to commit, and when to ship a pragmatic solution. You can balance immediate delivery with investments that make future work faster and more reliable.