Senior AI Agent Systems Engineer - Harness & Runtime (WW-PT)

Role Overview

We are seeking a highly skilled Senior AI Agent Systems Engineer to enhance the runtime and harness that power our AI agents. These agents are designed to operate across multiple platforms, executing real business tasks and working alongside human teams. Your primary focus will be on ensuring these agents are reliable, fast, and capable of continuous operation. This role is crucial for our next stage of growth, focusing on improving the quality, reliability, and responsiveness of our agent system.

Responsibilities

• Enhance systems managing user requests, model responses, and tool executions to improve reliability and performance.
• Eliminate issues such as silent stalls, duplicate responses, and lost work, ensuring accurate tool-status reporting.
• Develop resilient reconnect and recovery behaviors for long-running sessions.
• Design and improve context and memory management, ensuring agents retain important information without exceeding limits.
• Build tools centered around user intent, improving tool schemas, error messages, and execution reliability.
• Reduce latency and improve performance across all stages of an agent turn.
• Strengthen reliability across hundreds of persistent agent environments, ensuring robust failure recovery and message processing.
• Enhance product experience by improving React surfaces that display agent progress and tool activity.

Required Skills

• Proven experience in shipping production agent-runtime, LLM infrastructure, or autonomous workflow systems.
• Expertise in solving production harness problems, including agent loops, streaming management, and tool-execution frameworks.
• Strong proficiency in TypeScript and Node.js, with experience in building complex backend systems and APIs.
• In-depth understanding of distributed systems, message durability, retries, and state management.
• Ability to work independently within a small technical team and excellent written communication skills.
• At least six hours of working overlap with US Central Time.

Nice to Have

• Experience with coding agents, browser agents, or multi-agent platforms.
• Familiarity with multi-provider LLM routing and prompt caching strategies.
• Experience with Postgres or Supabase at scale and realtime systems using WebSockets or event streams.
• Knowledge of Docker fleet management and multi-channel messaging integrations like Telegram or Slack.
• Experience in building evaluation, observability, and debugging systems for AI agents.