Staff Machine Learning Engineer, AI Agent

Zendesk · Enterprise · San Francisco, CA +1

Staff Machine Learning Engineer focused on building AI Agent experiences for customer support, involving reasoning, tool use, and integration with Zendesk systems. The role emphasizes end-to-end product development, reliability, and technical leadership.

What you'd actually do

  1. Lead the design and development of AI Agent capabilities that can understand customer intent, follow conversation context, take reliable actions, and deliver helpful resolutions.
  2. Build agentic workflows that connect reasoning, retrieval, tool use, policy guidance, and human handoff into seamless customer support experiences.
  3. Partner with product, design, engineering, and research teams to turn customer problems into polished AI-powered product features.
  4. Improve the quality, reliability, and safety of AI Agent behavior through evaluation frameworks, feedback loops, observability, and iterative experimentation.
  5. Design systems that enable AI Agents to use Zendesk data, knowledge, business rules, and third-party integrations effectively and securely.

Skills

Required

  • Python
  • LLM-based applications
  • conversational systems
  • agent orchestration
  • retrieval systems
  • AI workflow automation
  • production-quality systems
  • APIs
  • distributed services
  • observability
  • testing
  • reliability practices
  • LLM evaluation
  • prompt and policy management
  • safety guardrails
  • tool calling
  • feedback-driven improvement loops

Nice to have

  • customer service
  • CRM
  • ticketing
  • knowledge management
  • enterprise workflow products
  • combining structured business data, unstructured knowledge, and real-time user context

What the JD emphasized

  • Deep experience building production AI, automation, or intelligent product experiences that solve real customer problems.
  • Experience designing agentic systems, conversational AI, workflow automation, retrieval-augmented experiences, or tool-using AI applications.
  • Ability to make complex AI systems reliable in production through evaluation, monitoring, debugging, and continuous improvement.
  • Hands-on experience with LLM-based applications, conversational systems, agent orchestration, retrieval systems, or AI workflow automation.
  • Familiarity with LLM evaluation, prompt and policy management, safety guardrails, tool calling, and feedback-driven improvement loops.

Other signals

  • AI Agent capabilities
  • agentic workflows
  • customer support experiences
  • AI-powered product features
  • AI Agent behavior
  • AI Agents to use Zendesk data
  • scalable agent architecture
  • AI-driven support