Katkat
Back

Design an Automated Ticket Investigation Agent

MetaSoftware Engineer · Technical Screen
HardSoftware EngineerTechnical ScreenML System Design

Last updated

This question evaluates a candidate's ability to design AI-enabled agentic systems for automated ticket investigation, covering competencies in ticket understanding, classification and prioritization, multi-source evidence retrieval (logs, metrics, traces, documentation), root-cause diagnosis, remediation orchestration, and human-in-the-loop safety controls. It is commonly asked in the ML system design domain to assess architectural and integration thinking, data retrieval and evaluation strategies, production monitoring and safety considerations, and tests both high-level conceptual architecture and practical application-level implementation concerns.

Design an AI-enabled agentic system that automatically investigates support or engineering tickets.

The system should:

  • Read an incoming ticket and understand the user's issue.
  • Classify and prioritize the ticket.
  • Gather evidence from internal tools such as logs, metrics, traces, documentation, runbooks, deployment history, and previous incidents.
  • Produce a diagnosis or likely root cause.
  • Recommend a resolution, escalation path, or next action.
  • Optionally execute safe remediation steps when allowed.
  • Keep humans in the loop for risky actions.

Discuss the agent loop, architecture, tool integrations, data retrieval, safety controls, evaluation strategy, and production monitoring.