Product Manager, Agent Harness & Modelling

Overview

In this role, you will own the North agent harness—the execution layer that makes agents reliable and production-ready. You’ll drive the agent loop, context management, and tooling orchestration to support long, multi-step workflows in enterprise settings. You’ll collaborate closely with engineering and the Modeling team to validate capabilities and evolve the harness in step with model progress. You’ll translate real-world enterprise feedback into concrete product and model requirements, shaping a scalable, secure agent platform for impactful automation.

Pay / Benefits

• health and dental benefits
• 6 weeks vacation
• remote-friendly with global office presence and stipends
• parential leave top-up
• co-working stipend
• weekly lunch stipend and snacks

Responsibilities

• Define and own the North harness roadmap across agent loop, context engineering, tool orchestration, sandbox execution, and sub-agent delegation
• Serve as primary interface between North engineering and Cohere’s Modeling team to validate new harness capabilities before build
• Own North’s agentic evaluation framework ensuring compatibility with training infrastructure and serving as a bridge between product and research
• Engage enterprise customers to surface agentic failures and translate findings into product and model requirements
• Stay current with open-source and commercial agent ecosystems to guide adoption and architecture alignment

Key requirements

• 5+ years of product management experience in agentic AI systems, developer infrastructure, or applied ML products
• Deep understanding of modern LLM agent architectures (multi-agent systems, tool-augmented reasoning, memory and retrieval, programmatic orchestration, RAG, long-horizon execution)
• Strong grasp of agentic evaluation design (measuring task completion, failure recovery, long-horizon reliability) and diagnosing model vs. scaffolding gaps
• Technical depth to contribute to architecture decisions at implementation level (design docs, async execution, filesystem design, sandboxed environments)
• Ability to switch between ML research discussions and engineering architecture conversations
• Track record of shipping platform-layer products with measurable impact on reliability, performance, or capability

• Cross-functional collaboration
• Strategic thinking and roadmap ownership
• Effective communication with engineering and research teams
• LLM agent architectures
• Multi-agent systems and tool-augmented reasoning
• Memory and retrieval mechanisms

…