Senior Software Engineer, Agent Infrastructure

Cohere Cohere · AI Frontier · Toronto, ON · Agentic Platform

Cohere is seeking a Senior Software Engineer to build agentic AI infrastructure, focusing on the platform that powers autonomous AI agents at scale. This role involves creating secure execution environments, managing agent state, routing models, and handling resource management for agent workflows. The ideal candidate has strong systems fundamentals, experience with ML infrastructure, and hands-on work with agentic systems.

What you'd actually do

  1. Turn emerging ML research ideas into production-ready infrastructure
  2. Build core platform capabilities for execution, storage, and state management
  3. Prototype and evaluate new technologies, then help decide what should move into production
  4. Partner with research teams to shape infrastructure based on what future agent systems will need

Skills

Required

  • Experience building production ML infrastructure with strong systems fundamentals
  • Hands-on work with agentic systems, multi-agent workflows, or agent development frameworks
  • Familiarity with model routing and LLM provider frameworks across different model types and environments
  • Experience with scalable, fault-tolerant distributed systems and Kubernetes
  • A track record of moving quickly on prototypes and making good decisions about productionization

Nice to have

  • Experience across on-prem, private cloud, and public cloud environments
  • Familiarity with storage systems, embedded databases, or filesystem abstractions
  • Experience with code execution sandboxes such as gVisor, Firecracker, Kata, or WASM runtimes
  • Interest in emerging ML infrastructure, edge inference, or browser-native models
  • Open-source contributions to LLM or agent infrastructure projects
  • Experience with identity, workload auth, or capability-based security systems

What the JD emphasized

  • Secure execution environments for agent-generated code
  • Identity, authentication, and trust boundaries for agents
  • Model routing and orchestration across different model types and environments
  • Rate limiting, quotas, and resource management for agent workflows
  • State management, memory, and filesystem abstractions for agents
  • Hands-on work with agentic systems, multi-agent workflows, or agent development frameworks
  • Familiarity with model routing and LLM provider frameworks across different model types and environments
  • A track record of moving quickly on prototypes and making good decisions about productionization

Other signals

  • building agentic AI infrastructure
  • platform that powers autonomous AI agents at scale
  • turning emerging ML research ideas into production-ready infrastructure
  • build core platform capabilities for execution, storage, and state management