Staff+ Software Engineer, Platform Ecosystem

Anthropic•San Francisco, CA

5h•$405,000 - $485,000•Hybrid

About The Position

The Auth & Identity team builds the authentication, authorization, and identity platform that underpins every Anthropic product: the API, Claude Code, Claude.ai, and our Enterprise offerings. Every login, API key, OAuth token, access decision, and organization record flows through the systems we own. As Claude moves from a chat product to an agentic platform, we are defining what identity and access look like when the actor is a human, a service account, or an autonomous agent acting on someone's behalf. We are a platform team with a product surface. The horizontal is the set of services every product consumes: credential minting and exchange, authorization controls, and the canonical record of users, service accounts, agents, and the organizations they belong to. The vertical is the experience where admins author that policy: SSO and SCIM setup, roles and permissions, IP allowlisting, and API key policies. Our work falls into four pillars: Authentication – Who you are. Credential minting, exchange, and lifecycle; sessions; workload identity federation; auth for async and agentic workloads. Authorization – What you can do. The policy framework for model, membership, role, and resource access. Identity – Who belongs to what. The tenant and organization model, and first-class identity for humans, service accounts, and agents, portable across surfaces and cloud environments. Enterprise – How enterprises connect. SSO, SCIM, login policies, and the admin controls that let customers provision and govern users from their own identity provider.

Requirements

8+ years of experience building and operating backend or platform systems at scale, ideally including identity, authentication, authorization, or security-adjacent infrastructure.
Worked with some of: OAuth 2.0 and OIDC, SAML, SCIM, JWTs and token exchange, workload identity federation, policy engines such as Cedar or OPA, or RBAC and resource-level access control models.
Care deeply about reliability and security. Design for on-call, failure modes, revocation, and the engineer who inherits your system, and treat availability of auth as table stakes for every product built on it.
Comfortable owning large migrations in live, high-traffic systems: shadow modes, incremental rollouts, and clean retirement of legacy paths.
Take a product-focused approach to platform work and build primitives that are easy for other engineers to adopt correctly, with documentation and self-service as part of the deliverable.
Can navigate ambiguity and make sound technical decisions independently, and communicate clearly with cross-functional partners and stakeholders.
Familiar with using AI tools as part of your software development workflow.

Nice To Haves

Experience with globally distributed databases and multi-region service architecture.
Experience working in Go, Python, Rust, or TypeScript across a rapidly changing codebase.
Prior work on enterprise identity integrations with major identity providers and cloud IAM systems.

Responsibilities

Own critical, high-traffic systems end to end, from design through rollout, operations, and iteration, and help shape the technical strategy for how identity and access work across Anthropic.
Designing credentials for agents and long-running autonomous workloads.
Routing every access decision through a unified authorization system, with low latency, security guarantees, and multi-region resilience.
Building a first-class identity model for service accounts and agents, and linking those identities to other identity and auth systems.
Global distribution and multi-region failover for some of the company’s most mission-critical services.
Deepening enterprise integration: login policies and recovery methods, self-service allowlisting, and tighter integrations with providers.
Work closely with product, security, infrastructure, and the product teams that consume our primitives.
Set technical direction, not just execute on it.