SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Anthropic is hiring a Staff Software Engineer to join the Agentic Systems team within the Platform organization. This role focuses on building and scaling Claude Managed Agents, a hosted platform for developing, running, and scaling production agents on Claude.
You'll own end-to-end systems from API design through production operations. Key responsibilities include:
**Scale the Platform**: Design and operate infrastructure for long-lived, stateful sessions that execute autonomously for minutes to days, persisting through disconnections and resuming cleanly. This spans Anthropic-hosted sandboxes, self-hosted customer environments, and multi-cloud deployments. You'll own durable session and event storage, sandbox orchestration, streaming, scheduling, and multi-tenant isolation—treating reliability, latency, and cost efficiency as product features.
**Evolve the Agent Harness**: Work alongside research to revisit harness assumptions with each Claude model generation. Build eval infrastructure that measures harness quality against research baselines and real customer workloads, ensuring the harness extracts maximum value from Claude.
**Enable Builders**: Ship capabilities that expand what agents can do—outcome-driven execution with success criteria and budgets, multi-agent orchestration, memory systems, and observability/tracing for long-running agents. The goal is highest intelligence per dollar on any agent platform, delivered safely.
**Design Durable APIs**: Shape primitives (agents, environments, sessions, vaults, event streams) that thousands of developers build against. Focus on versioning, ergonomics across API/SDK/CLI, sensible defaults, and escape hatches—with the expectation that implementations change while contracts hold.
You should have 8+ years as a backend, distributed systems, or infrastructure engineer. Experience building and operating stateful, long-running, or high-throughput systems (workflow orchestration, streaming, storage, container/job orchestration) is essential. Strong product sense, API design craft, comfort with 0→1 work and ambiguity, and hands-on use of Claude or AI tools are expected. You'll take full ownership from design through build, deployment, operations (including on-call), and iteration.
