Software Engineer, Sandbox & Agent Executor

Responsibilities

• Own agent behavior across product surfaces, including quality, safety, variance, failure modes, and tool and harness design.
• Design prompting, context construction, retrieval, routing, tool-use strategies, and evaluation systems for long-horizon workflows.
• Build and deploy agentic products in cloud and self-hosted environments, including sandboxed runtime infrastructure.
• Diagnose and resolve nondeterministic agent failures using transcripts, traces, statistical signals, and production evaluation data.
• Partner with product and infrastructure teams on provisioning, isolation, rollout, reliability, and safe deployment of agent workloads.
• Mentor engineers and establish standards for reliable agentic product development.

Requirements

• 6+ years of professional engineering experience owning complex production systems.
• Production experience shipping agentic systems with context engineering, tool use, and evaluation frameworks at real user scale.
• Hands-on experience with sandboxing technology and running agents in sandboxed environments.
• Practical experience with Kubernetes or equivalent platforms such as EKS or ECS, with end-to-end service ownership.
• Experience working across product, infrastructure, and evaluation layers of AI systems.
• Experience mentoring engineers on agentic systems and engineering judgment.
• Preferred experience building developer-facing surfaces such as CLIs, IDE extensions, or coding harnesses.
• Preferred experience shipping into enterprise or air-gapped environments.