Freelance Agent Evaluation Engineer at Mindrift

Mindrift connects experienced engineers with project-based AI work for top tech companies. We are currently seeking Freelance Agent Evaluation Engineers to help test, evaluate, and improve AI agents through structured, real-world scenarios. This role is ideal for engineers who enjoy flexible, project-based work and want to contribute directly to the development of cutting-edge AI systems.

Your Role

As a Freelance Agent Evaluation Engineer, you will:

• Design test cases and establish gold-standard evaluation criteria for AI agents
• Analyze agent behavior, logs, and failure modes to identify areas for improvement
• Build, iterate, and optimize prompts, evaluation scenarios, and testing logic
• Work with code repositories, test frameworks, and structured formats like JSON/YAML

Required Skills and Experience

To succeed in this role, candidates should have:

• 3+ years of software development experience, with strong Python skills
• Proficiency with Git, JSON/YAML, and Docker
• Understanding of LLM limitations, AI behavior, and evaluation design
• English proficiency at B2 level or higher

How It Works

• Freelance, project-based work with ≈6–10 hours per week during active phases
• Flexibility to choose projects and work on your own schedule
• Paid per project/task, with rates up to $80/hour depending on scope and expertise

This Freelance Agent Evaluation Engineer role is perfect for software engineers passionate about AI, testing, and iterative improvement, looking for flexible, high-impact work with leading technology teams.