Staff Backend Platform Engineer

COMPANY INFORMATION:

Our client is an AI-powered enterprise automation company that helps large organizations streamline complex operational workflows through intelligent AI agents. Its technology captures how employees perform tasks across multiple applications, transforms those activities into repeatable workflows, and enables AI agents to execute them autonomously, with human oversight and approvals when needed. Our client serves major enterprise customers, including U.S. financial institutions and a leading managed-cloud services provider.

JOB OVERVIEW:

Our Client runs a platform for enterprise AI agents with workflow management and auditing. The Staff Backend Platform Engineer focuses on the server-side foundation for observations, workflow execution, run scheduling and orchestration, approval gates, and clear audit trails. Work includes shaping the data model and external APIs, defining the run lifecycle from queueing through idempotent start and stop, and building recovery plus human-approval pause and resume. Core tech: Node with TypeScript, typically NestJS or a similar framework. PostgreSQL with heavy use of transactions, explicit locking, row-level security, and migrations applied while the system is online. MCP is part of the toolset, alongside OpenID Connect and Microsoft Entra SSO. There is also some Python.

Staff-level scope with ownership of core backend platforms, fully remote across the US. Work alongside engineers building agent runtimes and APIs, and collaborate with customer ops and IT on deployability and SSO. The product must run in a vendor cloud, a customer's cloud account, and on-prem inside customer networks, so portability and clear design docs matter. Worth a look if you care about audit, accountability, and safe automation at enterprise scale

JOB RESPONSIBILITIES:

• Ownership of the server-side platform: observations, workflows, runs, approvals, and the accountability record
• Data model and public APIs for observation and execution are yours to define and evolve
• Design the run lifecycle, including queueing, idempotent start and stop, recovery on worker disconnect, and human-approval pause and resume
• Concurrency safety across the system, preventing double-claim of queue items and making sure approvals reach the accountable person
• Treat approval policy as data: pre-approve vs wait, approver roles, and adaptive loosening based on successful history
• Full accountability across every action, tying activity to a person and recording what ran, stop points, approvals, and outcomes
• Tenant isolation for a multi-tenant product, including when a customer serves its own customers
• Design and own the externally facing APIs and the MCP tool surface that agents invoke
• Deployability across environments, the same product running in a vendor cloud, a customer cloud account, and on-prem inside a customer network
• Clear, strong design docs others can build from, and thoughtful reviews of implementations

REPRESENTATIVE PROJECTS:

• Reliable Run Initialization: Implement a fault-tolerant process that allows workflow runs to be safely retried after network failures without creating duplicate executions.
• Automated Worker Failover: Develop a lease-based mechanism that detects unresponsive workers and automatically transfers their active runs to available workers.
• Batch Approval Management: Enable authorized users to approve multiple similar workflow runs through a single action while maintaining individual audit records for each execution.
• Enterprise Audit and Compliance: Build a comprehensive, exportable audit trail that captures workflow executions, approval decisions, and approver identities to support banking compliance and regulatory requirements.
• Enterprise Authentication and Role-Based Access: Integrate with customers' existing identity providers to enable secure single sign-on (SSO) and role-based access controls (RBAC), restricting workflow execution and approval permissions to authorized users.

JOB REQUIREMENTS:

• 8+ yrs building backend systems
• Owned the data model for a multi-tenant product
• Expert-level TypeScript on Node, using NestJS or a comparable framework
• PostgreSQL mastery across transactions, locks, row-level security, and online migrations
• Have built long-running jobs and workflow systems yourself, with explicit state machines, retry logic, and idempotent APIs
• Delivered enterprise features such as permissions, audit logging, tenant isolation, and passed customer security reviews
• At least one shipped, user-dependent feature using an LLM or an agent
• Clear, strong writing and the habit of leading through design docs
• Consultative and collaborative with customers, able to sit with ops and IT and meet business stakeholders and operators
• Practical agent experience in real business processes such as ops, support, finance, or back office for non-developer users
• Daily personal use of agents in engineering work
• Able to articulate agent limits today, expected near-term changes, and failure modes around approvals, accountability, audit, permissions, legacy no-API systems, and human-in-loop
• Production agent experience beyond demos, not researcher-only and not prompt-engineer-only
• Temporal, LangGraph, or similar workflow and agent runtimes preferred
• OpenID Connect, Microsoft Entra SSO, and user provisioning preferred
• Python services experience preferred
• Experience delivering software deployable in customer networks as well as cloud services preferred