Help Outlier improve autonomous agents using agent debugging, stress testing, and technical feedback through remote freelance work.
This contributor opportunity supports AI organizations training large language models to perform proactive, multi-step tasks. You will evaluate complex architectural workflows and identify subtle failures in live interactions. The work suits software developers and orchestration specialists with two or more years of relevant experience.
Key Responsibilities:

Debug agent traces across multi-turn interactions
Stress test autonomous agents and workflows
Write clear evaluation rubrics for agent performance
Identify privacy leaks and authority escalation
Detect indirect prompt injection failures
Requirements:

Two or more years in backend engineering or AI automation
Experience integrating agents with live tools and APIs
Build modular, production-grade software systems
Handle persistent state and session discovery
Provide precise technical feedback on complex behavior
Bring agent evaluation, backend engineering, and systems integration skills to Outlier’s autonomous-agent training projects.

#J-18808-Ljbffr

AI Agent Evaluation Engineer Outlier