Claude is trained and served on hardware in data centers Anthropic operates and in sites partners build and run for us. DCSE secures that footprint: the sites, the hardware moving through them, and the people and vendors who touch it. You'll be the engineer who owns autonomous systems across that footprint as one security domain, from both sides: drones, ground robots and automated vehicles for patrol, inspection and moving hardware, including the security of that automation inside secure areas, and the detection systems that tell us what is flying over or operating near a site uninvited.
You set the approach, select and integrate the platforms and sensors, write the requirements vendors build to, and run what you ship, from the operating model where operators assign missions and machines carry them out to the playbook that turns a detection into an identified track and a well-evidenced handoff to whoever holds authority over the airspace. Interdiction is not part of this role. The goal is security that runs itself: coverage, fleet health and evidence verified continuously, with tools and AI agents doing the routine triage. We favor simple, standard controls and tools that can be verified over elaborate ones. We use Claude throughout our engineering workflow and expect you to as well.
You'll work with the rest of DCSE, including its physical security systems engineers and the engineers on its security monitoring platform, with Anthropic's data center operations and legal teams, and directly with site operators, equipment vendors and aviation and public safety authorities. The role is based in San Francisco, with remote applicants in the US considered, and involves regular travel to data center sites, which are critical infrastructure.
Key responsibilities
• Autonomous security and inspection systems: own the drones, ground robots and other autonomous platforms for perimeter patrol, incident response and infrastructure inspection: platform selection, operating model, operator qualification, maintenance, and the data path from machine to analyst
• Tools, automation and AI agents: build the integrations that carry detections, tracks, telemetry and fleet health into the team's security monitoring platform, and the tools and agents that triage detections, check fleet health and assemble evidence. When a review keeps costing people time, you build the tool or agent that retires it
• Detection of uncrewed systems: engineer the sensing that tells us what is flying over or operating near a site, what it is and who needs to know: sensor selection and evaluation against live targets, per-site coverage, classification and false-alarm handling, and the playbook from first detection to handoff, with evidence retained
• Security of site automation: engineer the controls on automated vehicles, robots and similar machines that move hardware or work inside secure areas: identity and access, network isolation, what they may record and where it goes, fail-secure behavior, their place in chain of custody, and acceptance before they enter a site
• Regulatory and site constraints: establish what we may sense, fly, drive, record and retain at each site, partner-operated sites included, under the aviation, spectrum, privacy and safety rules that apply there, with the waivers and approvals that extend it and coordination with local authorities set up before it is needed
• Health monitoring: define and test heartbeats, loss-of-coverage and lost-link alarms and fail-secure behavior, so a silent sensor or machine shows as a fault, not a quiet day
• Vendor requirements and acceptance: write the requirements, evaluate against live flights and runs, and hold vendors to what they ship
Minimum qualifications
• Have engineered autonomous or sensing systems in the field (drones, ground robots, automated vehicles, or RF, radar or optical detection sensors): selected, integrated or secured them for an operation other people relied on, and verified what they actually do against real targets or runs rather than a datasheet
• Have integrated systems you did not build (vendor APIs, SDKs, appliances, machine and sensor telemetry) into a platform that outlives any one of them
• Read and review code, design the data systems behind security tooling, and build automation (often with AI coding agents) using Git and code review
• Have done this work hands-on for data centers, critical infrastructure or a large hardware fleet
• Have designed inside a regulatory or safety constraint (aviation, spectrum, privacy, machine safety or similar) and can show the decision it drove
• Own a domain end to end: set the approach, carry it through with vendors, site operators, partners and authorities, and run it as something people depend on every day
