AI Systems Engineer

AppLovin•Palo Alto, CA

About The Position

AppLovin makes technologies that help businesses of every size connect to their ideal customers. The company provides end-to-end software and AI solutions for businesses to reach, monetize and grow their global audiences. For more information about AppLovin, visit: www.applovin.com. To deliver on this mission, our global team is composed of team members with life experiences, backgrounds, and perspectives that mirror our developers and customers around the world. At AppLovin, we are intentional about the team and culture we are building, seeking candidates who are outstanding in their own right and also demonstrate their support of others. Fortune recognizes AppLovin as one of the Best Workplaces in the Bay Area, and the company has been a Certified Great Place to Work for the last four years (2021-2024). Check out the rest of our awards HERE. We are building a layered AI intelligence system — a multi-layer agent architecture with a dynamic router, context engine, execution loop, verification layer, and eval feedback cycle — designed to handle long-horizon business tasks that cannot be accomplished in a single model inference. We are looking for engineers who understand that building with LLMs is not the same as building conventional software — and who find that difference interesting, not frustrating. You will own one or more layers of our AI system, from design through production, and iterate on them based on real-world failures. You will be matched to one of these layer areas based on your background and interest: Context engine: design and operate the RAG pipeline, memory architecture, MCP tool integrations, and prompt library that give the system company-specific knowledge at inference time Execution loop: build the action-observe-act cycle, tool integrations, state management, and logging infrastructure that let the system pursue multi-step goals Verification layer: design the checker model, confidence scoring, and human escalation logic that prevents the system from committing to bad outputs

Requirements

1–3 years of experience building real systems that use LLMs — not just calling an API, but designing context, handling failures, and shipping to users
Strong software engineering fundamentals: you write clean, testable code and you think about what happens when things break
Genuine curiosity about how LLMs behave: you have noticed patterns in how models succeed and fail, and you have opinions about why
Ability to move quickly and iterate: we are building in a new space and the path forward involves learning from production

Nice To Haves

Experience with vector databases, embedding models, or retrieval systems
Familiarity with agentic frameworks: LangChain, LlamaIndex, AutoGen, or similar
Experience designing prompts systematically — treating prompt design as an engineering discipline