Forward Deployed Engineer at Anthropic, what the role actually is.

It's a senior engineer on the Applied AI team who embeds with one of Anthropic's strategic enterprise customers to ship Claude into production. The deliverables named in the public listing are MCP servers, sub-agents, and agent skills, plus white-glove deployment support inside the customer's environment. The role is posted in six US cities (Boston, Chicago, NYC, Seattle, SF, DC) at $200,000 to $300,000 USD base, in Paris at €205,000 to €220,000 EUR, in London, and as a Federal Civilian variant in SF/NYC/DC. Verified 2026-05-08 against the greenhouse listing.

Same Applied AI team, different shape. The FDE embeds inside one customer's systems and ships production code; the Architect works across multiple customers and ships reference architectures, integration patterns, and enablement. As of 2026-05-08, Anthropic has four Forward Deployed Engineer postings (Applied AI US, Federal Civilian, London, Paris) plus one manager role, against eighteen Applied AI Architect postings spread across Commercial, Enterprise Tech, Federal Civilian, Government Technology, Industries, National Security, Partnerships, Public Sector, Security, Startups, and State and Local Government. If you want someone in your repo, you want the FDE. If you want a partner who attends executive briefings and co-designs your reference architecture, you want the Architect.

Three artifact types named explicitly on the JD: MCP servers (Model Context Protocol servers exposing customer tools and data sources to Claude), sub-agents (Claude agents specialized to one slice of the customer's workflow), and agent skills (capabilities the customer's main agent calls into). The artifacts run inside the customer's production environment. The leave-behind is Claude-shaped: the eval surface, the prompt scaffolding, the SDK calls, and the agent contract are all written against Anthropic's API and Anthropic's tool-use semantics.

$200,000 to $300,000 USD base for the Applied AI listing across the six US cities, per the greenhouse posting on 2026-05-08. The Paris listing posts €205,000 to €220,000 EUR. The London and Federal Civilian listings don't post a public salary band. Total compensation including equity is not on the public JD; market commentary places top-of-band Anthropic offers higher when equity is included, but $200k to $300k is what the listing actually says for cash base.

The Applied AI team's stated mandate is 'most strategic customers.' In practice that's enterprises with a substantial existing or planned commercial relationship with Anthropic. If you're a $50M ARR Series B with a Claude pilot, you're more likely to get an Applied AI Architect for a few hours per week than a Forward Deployed Engineer embedded full-time. If you're a Fortune 500 with an eight-figure inference commit and a regulated workflow, an FDE engagement is the right ask. If you're somewhere in between and the bottleneck is senior MLE capacity, you have other options including model-vendor-neutral studios that ship a similar artifact list.

Same role shape, public-sector deployment surface. The Federal Civilian listing is posted in SF, NYC, and DC. White-glove deployment support inside US federal civilian agency environments, with the same MCP servers / sub-agents / agent skills deliverable list. The compliance and clearance posture differ from the Applied AI track. Anthropic also posts 'Applied AI Architect, Federal Civilian' and 'Applied AI Architect, National Security' as separate non-FDE roles for that customer segment.

A model-vendor-neutral version of the same engagement. Senior engineer named in the scoping call lands in your repo on week 1. Week 2 prototype runs in your staging on an agreed rubric. Week 6 handoff includes the agent code, rubric.yaml, eval/cases.yaml, .github/workflows/pilot-gate.yml, and runbook.md, with the model provider chosen by you (Anthropic, OpenAI, Bedrock, Vertex, Azure OpenAI, or open-weight). The eval harness runs the same cases against any of them, so swapping models after handoff is a config change rather than a re-architecture. fde10x runs that engagement under a fixed fee with a refund-and-exit gate at calendar day 14.

Per the listing: 3+ years in a technical, customer-facing role (FDE or a software engineer with consulting experience). Production experience with LLMs including prompt engineering, agent development, and evaluation frameworks. Strong Python and ideally a second language like TypeScript or Java. Preferred: financial services or healthcare/life sciences background, enterprise IT systems experience. Former technical founders are explicitly encouraged to apply. The listed travel expectation is about 25 percent.