Terminal-Bench 3.0 Expert – Senior Technical & Domain Specialist

Experience: 6+ Years, depending on domain

Job Type: Contractor

Compensation: $300 per approved task

Domain: Coding | Machine Learning | Media | Systems | Security | Hardware | Operations/Finance | Science

Role Overview

We are looking for experienced Terminal-Bench 3.0 Experts to design, implement, test, and validate high-quality benchmark tasks across technical and scientific domains.

Prior hands-on experience working on Terminal-Bench 3.0 projects is mandatory. Candidates without previous Terminal-Bench experience will not be considered.

Key Responsibilities

- Design realistic, multi-step Terminal-Bench 3.0 tasks based on domain-specific workflows.
- Build reproducible, self-contained computational environments with pinned dependencies.
- Create datasets, configurations, simulations, parameters, or other task-specific inputs.
- Implement expert reference solutions using programming languages or domain-specific tools.
- Develop automated tests,



graders, and validation criteria.

Universal Requirements

- Previous hands-on Terminal-Bench 3.0 experience – Mandatory.
- Bachelor’s degree from a Top 100 institute/university in the country.
- Must meet the experience and educational requirements for at least one eligible domain.
- Robust analytical, technical, documentation, and problem-solving skills.
- Ability to independently handle task creation, implementation, testing, and validation.

Preferred Skills

- Experience evaluating AI coding agents or terminal agents.
- Experience creating benchmark tasks, automated graders, or evaluation pipelines.
- Research Software Engineering experience.
- Publications or open-source contributions involving computational science.
- Experience converting research or experimental workflows into reproducible computational packages.

📌 Terminal-Bench Expert | $300/task (India)
🏢 Codefeast
📍 India