Software Engineer, Identity Platform

OpenRouter is the leading AI routing and infrastructure layer that enterprises use to access, manage, and optimize the best large language models across providers—without lock-in, capacity constraints, or unnecessary cost. We power the most advanced AI teams in the world by giving them the flexibility to move fast, scale confidently, and stay future-proof as models evolve.

As enterprise adoption of AI accelerates, OpenRouter sits at the center of how organizations operationalize LLMs across research, product, and production workloads.

Every request that flows through OpenRouter starts with a question: who is this, and what are they allowed to do? Millions of developers, thousands of organizations and a growing fleet of autonomous agents reach us through API keys, browser sessions, OAuth apps, MCP clients and federated cloud workloads. The credential formats differ, but every one of them has to resolve to the same trustworthy answer.

We're hiring our first Software Engineer, Identity Platform to envision, build and own identity across OpenRouter's inference gateway, marketplace and rapidly growing suite of agentic products. You will define how people, organizations, workspaces, apps and agents authenticate, what they can access, and how that access is delegated, governed and revoked, from a solo developer's first sign-in to a Fortune 500 rollout provisioned straight from their identity provider. If you love the intersection of security, systems design and product, this is your chance to build the trust layer that takes OpenRouter into the enterprise.

• Unify authentication. API keys, management keys, browser sessions, OAuth and OIDC tokens and MCP logins all resolve to a single principal with one authorization model. Own sign-in, MFA, passkeys and enterprise SSO (SAML and OIDC) on top of it, plus the full lifecycle of every credential we issue: scoped, expiring, rotatable and revocable.

• Build delegated access for apps and agents. Sign in with OpenRouter, OAuth for MCP clients with dynamic client registration, and consent grants that can be revoked without touching the underlying key. Apps and agents act on a user's behalf without ever handling a secret, and admins can pin that consent to an org or workspace across a managed fleet.

• Go secretless with workload identity. CI pipelines and cloud workloads exchange their own identity provider tokens for short-lived OpenRouter credentials under org-defined federation policies, and internal services authenticate to each other the same way. No static secrets to leak or rotate.

• Build composable authorization. A single permission model where every route asks "can this caller do this, here?" instead of checking role names. Org and workspace roles, viewer and restricted roles, read-only automation keys and, over time, custom roles. Deny by default, enforced on the server and reflected cleanly in the UI.

• Make enterprise provisioning just work. SCIM user and group provisioning, group-to-role mapping, domain verification and self-serve SSO that works first try with Okta, Entra ID, Google Workspace and the long tail of identity providers. Partner platforms can provision accounts and keys for their own customers programmatically. An IT admin should be able to onboard thousands of seats without filing a ticket.

• Define the core identity model. Users, organizations, workspaces, memberships, apps and guardrails, and how they compose. Clean boundaries, clear ownership semantics and a data model that supports the next five years of enterprise features without a rewrite.

• Protect users and their data. Identity fraud detection, KYC and verification flows, step-up checks on sensitive actions, account-takeover defenses and abuse signals that stop bad actors without adding friction for good ones. Least-privilege access to customer data across every OpenRouter surface.

• Treat auditability and correctness as features. The system ensures every credential has a provenance, every access decision is explainable and every privileged change lands in an audit log. Auth stays fast and highly available on the hot path of every inference request, with clear, standards-based failure modes developers can debug on their own. An access grant nobody can explain is a bug.

• Work cross-functionally. Partner closely with the Trust & Safety team on fraud and abuse, with the Money team on payment-linked verification and risk, with Product on dashboard and onboarding experiences, and with Sales and CS on enterprise security reviews and SSO rollouts.

• 5+ years building production identity, authentication or authorization systems at a SaaS, developer-platform or security company.

• Deep fluency in identity standards: OAuth 2.x (PKCE, token exchange, dynamic client registration), OIDC, SAML, SCIM and JWTs, and the ways they break in practice.

• Experience designing authorization models (RBAC, ABAC or ReBAC) for multi-tenant B2B products, including migrating a live system from ad-hoc checks to a unified model.

• Hands-on experience integrating enterprise identity providers (Okta, Entra ID, Google Workspace) and an auth platform (Clerk, Auth0, WorkOS or similar).

• Strong TypeScript and SQL. Comfortable with Postgres and building low-latency, cache-heavy services on a critical request path.

• A security mindset. You think in threat models, design for least privilege and can explain your decisions to an enterprise security reviewer.

• You work well across teams. You'll partner daily with Trust & Safety, Money, Product, Sales and CS, and turn messy enterprise asks into clean systems.

• High agency and AI-forward. You use coding agents daily and like automating away manual security and onboarding work.

• You've shipped enterprise SSO and SCIM that passed real customer security reviews.

• You've built workload identity federation or secretless auth for CI and cloud workloads.

• You've built OAuth, consent and delegation for AI agents or MCP clients acting on behalf of users.

• Experience with identity verification, KYC or fraud and risk systems (e.g. Stripe Identity, Persona, Sift).

• Experience with policy engines or fine-grained authorization (e.g. OpenFGA, Cedar, OPA, Zanzibar-style systems).

• You've helped a company achieve SOC 2, ISO 27001 or similar compliance.

• You're an existing OpenRouter user, or have built projects in AI tooling.

If you don’t think you meet all of the criteria below but still are interested in the job, please apply. Nobody checks every box, and we’re looking for someone who is excited to join the team.