AI Platform Engineer

Company Overview: Cellebrite’s (Nasdaq: CLBT) mission is to enable its global customers to protect and save lives by enhancing digital investigations and intelligence gathering to accelerate justice in communities around the world. Cellebrite’s AI-powered Digital Investigation Platform enables customers to lawfully access, collect, analyze and share digital evidence in legally sanctioned investigations while preserving data privacy. Thousands of public safety organizations, intelligence agencies and businesses rely on Cellebrite’s digital forensic and investigative solutions—available via cloud, on-premises and hybrid deployments—to close cases faster and safeguard communities. To learn more, visit us at www.cellebrite.com ,  https://investors.cellebrite.com/investors and find us on social media @Cellebrite. Position Overview As an AI Platform Engineer, you build the software that helps our engineers build our products. Everything the team ships is measured against one thing: whether it makes the development lifecycle faster and better, across planning, building, reviewing, testing, shipping and maintaining. This is a DevEx role that covers the full span, from the infrastructure our AI agents and automations run on to the surface developers touch every day. It does not sit at one end. It asks for someone deep enough to build the platform layer and hands-on enough to build the tooling on top of it. If you are strong on one side and credible on the other, that is a real candidate, not a compromise. You work inside R&D groups, find the problems worth solving, and build the solution together with the people who have the problem. The AI ecosystem changes every quarter. A large part of the job is staying current with it hands-on, deciding what is worth adopting, and right-sizing the answer. Sometimes that is a repo of markdown, sometimes a full agent pipeline. Key Responsibilities • Own the surface developers use every day: agents, skills, plugins and workflow • Own the infrastructure those run on, including how they are built, deployed, operated and governed • Build and maintain CI/CD, security gates, evaluations and observability for AI-based tooling • Build integrations across GitHub, Jira, Slack, CI/CD and whatever internal system the work needs next • Partner with engineering groups to turn a real pain point into something that works and holds up in daily use • Shape the architecture of the platform. The direction is set; the detail and the calls are yours • Run POCs to settle build-vs-buy questions with evidence rather than opinion • Design for token economy: cost, latency, context design and model choice • Drive adoption of what you build, and measure its impact on the development lifecycle

Requirements: • Strong in a modern backend language. Most of the work here is Python and TypeScript; if your depth is C++, C# or Java, you will pick those up here, and that is normal, not a concession. • Cloud, containers and CI/CD, and ease with tooling that keeps moving. • Strong integration skills: APIs, webhooks, authentication and secrets (OAuth, SSO, service accounts, token lifecycle), and the discipline to build things other people can rely on. • Hands-on with the current AI platform stack: LLMs, MCP and tool design, agent orchestration and subagents, RAG (hybrid retrieval, reranking, knowledge graphs), context budgets and prompt caching, model routing across providers. The field moves monthly, so we expect gaps, and we care that you know where yours are. • Real platform and backend depth, with production ownership. You have built systems that broke in production, and you fixed them. • You have built something real with AI, not just used the tools. Something that ran outside your own machine and that other people used: a skill or agent wired into CI, an internal tool with actual users, an automation a team came to depend on. You can say how it behaved in practice, where it broke, and roughly what it cost to run. • You treat non-determinism as an engineering problem rather than a caveat: evals and regression suites including LLM-as-judge, structured outputs and reliable tool calling, guardrails against prompt injection and tool misuse, sandboxing and least privilege for agents, per-run tracing of latency, tokens and cost. • Ability to work directly with engineering groups, identify high-value problems and deliver with them.