Senior Bash Code Reviewer — Shell Script Audit

About OpenTrain AI

OpenTrain AI builds careers in AI training and data labeling. We connect people with flexible, remote work that directly shapes how state-of-the-art AI behaves and grows.

As the hiring organization for this role, OpenTrain AI supports contributors with clear rubrics, quality standards, and a platform to apply, learn, and progress.

About AI training (data labeling) for code

AI training — also called data labeling or annotation — is the human process that teaches models how to write, evaluate, and improve code. For programming tasks, human reviewers validate outputs, correct mistakes, and provide feedback so models learn safe, reliable patterns.

This role focuses on Bash and POSIX shell scripts: your reviews and corrections will be used to train next-generation models that generate and audit shell code.

The role

You will audit annotator reviews of AI-generated shell scripts, executing and validating submissions in sandboxed environments and ensuring each script meets the prompt and safety standards.

Your corrections and feedback will enforce a quality rubric covering correctness, cross-shell compatibility, security, error handling, and idempotence.

• Validate scripts run correctly across POSIX-compliant shells and common Bash versions (4.x/5.x).
• Identify mis-ratings and provide clear, actionable feedback to bring reviews up to rubric standards.
• Execute proof-of-work tests in isolated VMs or containers to confirm compliance with prompts.

What you'll do day-to-day

Work through annotated submissions, run scripts safely in sandboxes, reproduce issues, and confirm whether the code satisfies the original prompt and tests.

Apply security and reliability best practices when judging scripts; correct unsafe ratings and explain fixes so annotators learn and the dataset improves.

• Run and test scripts in isolated environments (containers or VMs) to validate behavior and side effects.
• Check for POSIX compatibility, quoting and escaping, input sanitization, least-privilege file handling, and absence of unsafe eval/sourcing.
• Assess error handling and idempotence: use of set -euo pipefail, meaningful exit codes, and re-runnable logic.
• Use tooling such as shellcheck, bats, and bash -x tracing to diagnose and document issues.
• Score reviews against a rubric and leave concise, constructive corrections and mentoring notes for annotators.

Requirements

We seek a seasoned Bash reviewer who can reliably validate code, spot security and reliability issues, and communicate corrective feedback to annotators.

• Experience: 5–7+ years in professional Bash scripting, Linux systems administration, or code-review roles.
• Shell mastery: Expert knowledge of Bash 4.x/5.x and POSIX sh compatibility (subshells, pipelines, traps, env config).
• Secure scripting: Skilled at finding unescaped variables, injection risks, unsafe file handling, and unsafe eval/sourcing.
• Error handling & idempotence: Proven experience with set -euo pipefail, exit codes, and re-runnable logic.
• Testing & debugging: Proficient with Bats, shellcheck, and bash -x tracing.
• Platform integration: Familiarity with cron jobs, systemd units, and containerized Linux environments.
• Proof-of-work: Comfortable executing scripts in isolated VMs/containers to verify prompt compliance.
• QA practice: Experience with rubric-based scoring, checklist reviews, and ticketing/annotation tools (Jira, Asana).
• Communication: Excellent written English (B2+ CEFR), able to deliver clear, constructive feedback and mentor peers.
• Nice to have: Exposure to LLM evaluation, RLHF pipelines, or prior AI/ML data-labeling projects.

How this role works

This is a remote contractor, part-time role. You will work 20+ hours per week and follow OpenTrain AI’s review rubrics and QA processes.

Compensation is pay-per-hour at USD 21/hour. Work is worldwide remote; you will execute code only in isolated, provided sandboxes or your approved test environments.

• Employment types: Contractor, Part-time.
• Time: 20+ hours/week minimum commitment.
• Pay: USD 21 per hour (pay-per-hour).
• Apply: Create an OpenTrain account (free) and submit your application and profile through the OpenTrain platform.