Remote Freelance Agent Evaluation Engineer Job in Kuwait | Mindrift

Mindrift is recruiting for Freelance Agent Evaluation Engineer in Kuwait City, Kuwait. This page preserves the source-backed responsibilities, requirements, work details and current application route for the role.

Role overview

Part-time remote freelance work on Mindrift’s Tendem project, designing realistic software-agent evaluation environments and tasks that test whether AI coding agents can handle real developer workflows. The assignment is about constructing and evaluating realistic software-engineering tasks; it is not ordinary data labeling, prompt-only work or greenfield coding from scratch.

Responsibilities

• Build realistic developer environments containing a codebase, infrastructure, tickets, documentation and team-style context.
• Design intermediate-state tasks, prompts and clear solved criteria that reflect genuine engineering work rather than artificial toy problems.
• Write robust tests that accept multiple valid implementation approaches while rejecting incorrect outcomes.
• Iterate on tasks and environments based on QA feedback and evaluation results.

Requirements and qualifications

• At least five years of software-development experience is sought for this senior freelance work.
• Strong practical experience with technologies such as Python/FastAPI, JavaScript or TypeScript/React, Docker, PostgreSQL, Kafka and Redis is relevant.
• Experience designing functional and integration tests and reasoning about real software systems is required.
• English proficiency of at least B2 is requested; the source also references a relevant master’s degree as preferred, while a bachelor’s degree can be accepted with sufficient experience.
• Applicants are asked to submit an English CV and state their English level.

Benefits and work environment

• Part-time remote freelance work designed to fit around primary professional or academic commitments.
• Exposure to advanced AI projects that can strengthen the specialist’s portfolio.
• An opportunity to influence how future AI coding models are evaluated and improved.

Working time and compensation

• Published compensation: Up to about USD 50/hour equivalent per accepted task, depending on tier and pace.

How to Apply

Use the current job/application page below to submit your application or continue to the employer's active application flow.

Apply for this job

Prepare an up-to-date CV and make sure the information you submit matches the role requirements and location shown above.