AI Harness Engineer

Core Objectives

• Agentic Infrastructure: Build the foundational infrastructure, robust APIs, and secure tool-calling environments necessary for multi-step agent execution.
• Orchestration & State Management: Manage complex context windows, system prompts, persistent memory layers, and error-recovery loops to ensure long-running agent workflows complete reliably.
• Safety, Governance & Guardrails: Implement strict security permissions, deterministic input/output validation layers, and evaluation frameworks to monitor agent behavior and eliminate unauthorized actions.

Requirements

Key Responsibilities

• Design and engineer the repeating agent loop (model execution, tool parsing, reasoning steps, and termination logic).
• Optimize context window packaging, compaction strategies, and prompt caching to reduce latency and inference costs.
• Build scalable backend components and distributed microservices to support high-concurrency AI workloads.
• Implement robust monitoring, observability, and evaluation frameworks to measure LLM performance and agent reliability in production.
• Collaborate with data and security teams to build graph databases or vector storage systems for relationship reasoning and long-term agent memory.