Anthropic Applied AI and FDE roles: verified vacancies and interview guidance

Choose a vacancy before choosing practice questions

Tokyo has an Applied AI Engineer, an explicitly pre-sales Applied AI Architect and a partner-facing Architect. The verified FDE vacancies are in Europe. A past FDE role is an eligible background for Tokyo Engineer; that does not rename the Tokyo vacancy or make its interview identical to an FDE interview. This article is the factual reference for preparation, checked 2026-10-05. Its numbered facts are official statements or dated page observations; later practice articles identify their own inferences.

The evidence set contains 74 facts from 16 distinct official sources. An available application is a recruiting-state observation at the check date, not a promise that hiring remains open. Initial publication dates remain unknown; the AI policy additionally exposes a last-update date of 2025-07-10. No dated change in the rolling 2026-09-05–2026-10-05 window (Asia/Tokyo) was established from these sources; undated vacancies are not recent announcements. No interview anecdotes, private exam material or copied imagery are used.

Role inventory: Japan first, then regional comparisons

All rows were checked 2026-10-05. “Available” means the official application page was present. Internal levels are unknown; Manager explicitly denotes people management. Years of experience do not establish a level equivalence.

Official role and region	Job ID / observed state	Reading scope
Applied AI Engineer — Tokyo, Japan	5390799008 / available	Implementation + technical advice; not named FDE
Applied AI Architect — Tokyo, Japan	5076109008 / available	Pre-sales; covers discovery through deployment advice
Applied AI Architects, Partner — Tokyo, Japan	5439499008 / available	Pre-sales partner capability and indirect revenue
Forward Deployed Engineer — London, UK	5423029008 / available	Customer-system production delivery; 4+/8+ conflict
Forward Deployed Engineer — Munich, Germany	5391016008 / available	Regional German language and experience conditions
Forward Deployed Engineer — Paris, France	5391021008 / available	Regional French language and experience conditions
Applied AI Engineer — Sydney, Australia	5248983008 / available	8+ years; workshops and code review
Applied AI Architect — Sydney, Australia	5117589008 / available	10+ years; do not transplant to Tokyo
Applied AI Engineer, Enterprise Tech — San Francisco / New York City / Seattle, US	5057647008 / available	Enterprise technology portfolio; 3 office days/week
Manager, Forward Deployed Engineering — New York City, US	5099753008 / available	10+ technical / 2+ management years; quality and staffing
Technical Deployment Lead — Austin / Boston / New York City / San Francisco / Seattle, US	5017903008 / available	SOW, scope, stakeholders and value; no production coding
Pre-Sales Program Lead, Forward Deployed Engineering — San Francisco / New York City, US	5391012008 / available	Selection to contracting and delivery handoff
Former Forward Deployed Engineer URL — US shown only in old search result	5302966008 / unavailable	Availability check only; not a current US job

Jobs | Anthropic — Official index observation for the listed Applied AI vacancies.

Fact ID	Verified statement
A016	The index lists Tokyo Engineer/Architect/Partner roles and European FDE roles separately.

Former Forward Deployed Engineer URL — Observed unavailable job description and redirected destination.

Fact ID	Verified statement
A017	The old US FDE URL redirects to an error-marked index without a job description.

No Tokyo/APAC position titled “Forward Deployed Engineer” was found in the reviewed index. This is a search boundary, not a statement that the company has no such organization, future vacancy or unpublished role.

What the official hiring guidance actually establishes

Careers | Anthropic — General hiring logistics, principles, accommodations and reapplication.

Fact ID	Verified statement
A001	Interviews use Google Meet; timezone accommodation is available.
A002	Technical live-coding tool examples are Colab and CodeSignal.
A003	Lookup is allowed; know basic syntax and standard libraries.
A004	Experience, motivation and candidate questions are discussed.
A005	Around half of technical staff had no prior ML experience.
A006	Company principle: empirical work, simple functioning solutions, iteration.
A007	Reliability, safety, security and mission are shared principles.
A008	Accommodation requests are confidential; detailed medical information is unnecessary.
A009	Reapply after 12 months, or sooner following material experience/skill changes.

The general page does not publish an FDE loop. Google Meet and coding-tool examples establish logistics; they do not establish round count, mandatory coding for every role, customer simulation or scoring. Basic lookup permission does not authorize AI assistance.

A dated policy changes how preparation and assessment are separated

How to collaborate with Claude during our hiring process — Candidate tool use before and during assessment, and company hiring use.

Fact ID	Verified statement
A010	Write the first application draft yourself; AI refinement is permitted.
A011	AI may support company research, answer rehearsal and candidate questions.
A012	Take-home assessments exclude AI unless explicitly permitted.
A013	Live interviews exclude AI unless explicitly permitted.
A014	Show authentic experience and thinking; follow the policy transparently.
A015	Claude supports hiring; candidate data does not train Claude and Claude does not decide hiring.

The 2025-07-10 update permits AI during preparation and refinement, while defaulting to independent work during assessments. At the 2026-10-05 check this distinction remains visible. The candidate therefore needs two routines: prepare with feedback, then rehearse explaining and implementing unaided. If an assessment explicitly permits AI, follow that assessment's scope; do not infer permission from ordinary workplace use of Claude. The policy page exposes an update date, not a verified initial publication date.

Tokyo Engineer: implementation evidence plus customer judgment

Applied AI Engineer — Tokyo implementation duties, eligibility and visible application fields.

Fact ID	Verified statement
A018	Build Japan-enterprise LLM solutions meeting business, technical, safety and reliability needs.
A019	Show Claude value to prospects using architectures and prompting.
A020	Work with AEs on requirements, pilots, prototypes and evaluation suites.
A021	Advise integration and blockers through onboarding and beyond go-live.
A022	Relevant TPM/FDE/platform experience and strong Python proficiency are minimum qualifications.
A023	Native-level Japanese and business-level English are required.
A024	4+ relevant years and recent production-LLM experience are preferred.
A025	Preferred: ambiguity, understandable solutions and cross-team trade-offs.
A026	Customer visits are occasional; general office attendance is at least 25%.
A027	Application asks client experience, Python and motivation; suggested motivation length: 200–400 words.

Tokyo Architect: pre-sales ownership with technical depth

Applied AI Architect — Tokyo pre-sales responsibilities and qualification distinctions.

Fact ID	Verified statement
A028	Tokyo Architect is explicitly pre-sales.
A029	Supports customers using Claude API and Claude for Work.
A030	Advises discovery, evaluation, deployment and use-case evaluation frameworks.
A031	Communicates technical depth to engineers and business value to executives.
A032	Returns integration-pattern insights to Product and Engineering.
A033	Minimum: customer-facing technical experience and complex enterprise buying cycles.
A034	Native-level Japanese and business-level English are required.
A035	5+ years, cloud/enterprise integration, Python and LLM knowledge are preferred.
A036	Experience requirements correlate with internal levels; no internal level name is disclosed.

Tokyo Partner Architect: strengthen the delivery channel

Applied AI Architects, Partner — Tokyo partner responsibilities and preferred candidate background.

Fact ID	Verified statement
A037	Pre-sales technical relationships with GSIs/RSIs and AWS/GCP partners.
A038	Supports indirect revenue, partner AI practices and long-term go-to-market.
A039	Co-develops industry solutions, reference architectures and deployment practices.
A040	Unblocks strategic customer deals primarily delivered by partners.
A041	Partner events, developer enablement and field feedback are responsibilities.
A042	5+ customer/partner technical years, partnerships and fluent Japanese/English are preferred.

European FDE: embedded production work and unresolved form differences

London: the following eight facts are London-specific

Forward Deployed Engineer — London production delivery, travel, skills and body/form discrepancy.

Fact ID	Verified statement
A043	Embeds with strategic customers to build production AI within their systems.
A044	Works with Post-Sales, Product and Engineering on workflows, safety and reliability.
A045	Builds production MCP servers, sub-agents and agent skills.
A046	Codifies reusable deployment patterns and returns product/engineering insights.
A047	Customer-site travel is 25–50%.
A048	Seeks production LLM, prompting, agents, evaluations, scale and Python skills.
A049	Body says 4+ customer-facing years; application form asks 8+.
A050	Application asks enterprise production-LLM and embedded implementation experience.

Munich: regional conditions

Forward Deployed Engineer — Munich regional eligibility and public application wording.

Fact ID	Verified statement
A051	Munich lists 8+ customer-facing years, native German and fluent English.
A052	Munich form asks German C1+ and on-site production-LLM experience.

Paris: regional conditions

Forward Deployed Engineer — Paris regional requirements read from the English-language portion.

Fact ID	Verified statement
A053	Paris English section lists 8+ years, native French and fluent English.

London's 4+/8+ discrepancy and Munich's native/C1 wording remain unresolved. Neither the more permissive body nor the stricter form is silently substituted for the other. Use the job ID when asking which condition applies. Travel percentages describe time at customers; office percentages describe office attendance.

APAC and US comparisons: related work, distinct conditions

Applied AI Engineer — Sydney engineering responsibilities and candidate background.

Fact ID	Verified statement
A054	Sydney Engineer supports API architecture, evaluations and pilots.
A055	Runs hands-on customer-engineering workshops and code reviews.
A056	Sydney lists 8+ technical years, production LLMs, Python and agent/retrieval frameworks.
A057	Conference, blog and white-paper knowledge sharing is included.

Applied AI Architect — Sydney Architect experience requirement as listed.

Fact ID	Verified statement
A058	Sydney Architect lists 10+ customer-facing technical years.

Applied AI Engineer, Enterprise Tech — US enterprise technology responsibilities, skills and role-specific attendance.

Fact ID	Verified statement
A059	US Enterprise Tech Engineer supports Claude API adoption in customers’ core products with AEs and Architects.
A060	Examples include transcript analysis and MCP; languages include Python or TypeScript.
A061	Packages prompting, evaluation, agent and architecture knowledge into reusable assets.
A062	Specifies three office days per week.

Managers and delivery leads change the unit of responsibility

Manager, Forward Deployed Engineering — New York management ownership, experience and collaboration boundary.

Fact ID	Verified statement
A063	NYC player-coach owns hiring, development, quality and organization building.
A064	Owns staffing, architecture/code review and reusable delivery assets.
A065	Lists 10+ technical years, 2+ management years and zero-to-one organization building.
A066	Engagement Managers handle logistics/stakeholders; FDE Manager handles quality and growth.

Technical Deployment Lead — US delivery ownership, success measurement and technical responsibility split.

Fact ID	Verified statement
A067	TDL owns SOW-to-production delivery; FDEs build the technical solution.
A068	Owns scope, milestones, dependencies, success criteria, MVP and backlog.
A069	Sets value hypotheses, baselines and KPIs; measures pre/post-deployment ROI.
A070	Navigates security review, legal approval, procurement and compliance.
A071	Does not write production code; owns technical direction alongside FDEs.

Pre-Sales Program Lead, Forward Deployed Engineering — US pre-contract programs and subsequent delivery handoff.

Fact ID	Verified statement
A072	Pre-Sales Program Lead builds selection-to-contract processes.
A073	Balances commercial, research and mission goals; coordinates legal/finance contracting.
A074	Creates post-contract handoff and partner-delegation/co-delivery decisions.

Read the delivery boundary, not a presumed hiring loop

1. 1Tokyo Engineer
2. 2pilot, evaluation suite, integration
3. 3advice through and beyond go-live

1. 1Tokyo Architect
2. 2discovery, architecture, evaluation
3. 3enterprise adoption advice

1. 1Tokyo Partner Architect
2. 2partner capability, joint solution
3. 3partner-led customer delivery

1. 1US Pre-Sales Program Lead
2. 2selection and contract
3. 3post-contract delivery handoff

1. 1US Technical Deployment Lead
2. 2scope, milestones, stakeholders, ROI
3. 3FDE technical implementation

1. 1US FDE Manager
2. 2staffing, review, coaching
3. 3team quality and growth

Original responsibility map, supported by A018–A042 and A063–A074. Arrows show outputs and collaboration, not a formal organization chart, mandatory sequence or shared recruiting process. Read each row separately: the US lead/FDE split must not be imposed on every Tokyo engagement. The practical implication is to choose an experience that demonstrates the intended role's ownership. An implementation story does not by itself establish partner enablement or people management.

Build an evidence worksheet without inventing achievements

This is an original offline exercise, not a company assessment. Select one vacancy and two projects you actually worked on. For each project record the fields below; leave unknown values blank rather than replacing them with attractive numbers.

Field	Evidence to prepare	Decision it lets you explain
Your authority	Your code, review, decision or agreement; other owners named	What you owned and what you escalated
Workflow	User, input, output, existing process and failure consequence	Why the intervention belonged there
Evaluation	Case selection, baseline, denominator, period, bad-case categories	Whether improvement was repeatable and meaningful
Implementation	Interface, deployment boundary, logs, recovery and handoff	How a pilot became supportable
Customer delivery	Scope change, dependency, sponsor decision and outcome	Which constraint drove the trade-off
Reuse or teaching	Artifact, intended users, adoption and maintenance owner	Whether knowledge survived the engagement

Link each row to a relevant fact ID, then write one paragraph separating recorded result, your contribution, uncertainty and next decision. If you only prototyped, say so. If you lack production evidence, identify the gap; a synthetic exercise may demonstrate reasoning but cannot become past customer experience. Review technical controls with production AI controls.

Questions that remain for the recruiter

• No role-specific FDE or Tokyo Engineer round count, order, duration, scorecard or take-home requirement was established.
• Which roles require coding, system design or customer simulation is unknown; the careers page is general technical guidance.
• No internal question, algorithm list or pass threshold was established.
• Tokyo/APAC FDE titles were not found in this official-index check; nonpublic or future positions remain possible.
• London 4+/8+ and Munich native/C1 differences remain unresolved.
• Experience-year differences do not establish internal level, compensation or promotion equivalence.
• Specific data residency, retention, privacy or contract rules are not established by these vacancies.
• Partner revenue duties do not establish an FDE IC quota or commission scheme.
• The AI policy was last updated 2025-07-10 and checked 2026-10-05; confirm any newer policy and assessment-specific instructions.