About OpenTrain

OpenTrain AI is the hiring and contracting organization for this role and the #1 platform for finding and building careers in AI training and data labeling. OpenTrain connects skilled contributors with projects that help shape how modern AI systems learn, reason, and work with software.

This is a worldwide contractor opportunity. Creating an OpenTrain account is free, and candidates can apply in minutes.

Contractor engagement
Worldwide opportunity
Expected project duration of 6+ months
Payment of $3 per verified trace

About AI Training and Code-Agent Data

AI training is the human side of building artificial intelligence. For coding systems, experienced software engineers create and review examples of real development work so models can better navigate repositories, write and improve code, use terminal tools, debug issues, and validate results.

Your work will contribute authentic supervised fine-tuning trajectory data from real software development sessions. These detailed records help AI coding agents learn from realistic human-agent interactions rather than isolated code snippets.

Work on cutting-edge AI training focused on software development
Use authentic development workflows as the source of training data
Help improve how AI coding agents handle practical engineering tasks

The Role

OpenTrain is seeking expert software engineering contributors who are daily users of the specified code-agent CLI. You will capture complete, authentic development sessions and submit verified JSON or JSONL execution traces for simple, medium, and complex tasks.

Each trace must preserve the full session context and include the information needed to verify the work. The engagement includes an initial batch of 20 to 50 traces, weekly throughput expectations by complexity tier, pricing per verified trace, and sample logs when available.

Role: Code CLI Execution Trace Contributor
Experience level: Expert
Payment type: Per verified label or trace
Rate: $3 USD per verified trace
Expected duration: 6+ months

What You'll Do

You will document real software development sessions in a way that is complete, accurate, and independently verifiable. The submitted trajectory data should show the development task, the tools and terminal interactions used, the intermediate reasoning, and evidence that the resulting work is correct.

Contributors should be prepared to work across common programming stacks and to handle the full development lifecycle represented in each assigned task.

Submit verified JSON or JSONL execution traces
Complete simple, medium, and complex software development tasks
Provide the exact starting repository commit
Include root-level project guidelines and terminal or tool outputs
Preserve complete session context and authentic human-agent interaction
Document intermediate reasoning within the required trace format
Provide validation evidence such as passing tests and a clean git diff
Deliver an initial batch of 20 to 50 traces
Meet weekly throughput expectations by complexity tier

Requirements

This role is intended for experienced software engineers who can explain their relevant work clearly and maintain a high standard of technical accuracy. Strong attention to detail is essential because every trace must preserve context and provide reliable evidence of the outcome.

Fluent English proficiency is required. Contributors must also be able to discuss their experience with real code-agent CLI workflows and explain screening decisions clearly.

Daily user of the specified code-agent CLI
At least two years of professional software engineering experience
Practical expertise across common programming stacks
Experience with repository navigation, debugging, feature development, refactoring, and testing
Strong terminal workflow skills
Relevant software engineering and AI coding agent experience
Ability to preserve complete session context
Ability to provide reliable validation evidence
Strong accuracy and attention to detail
Ability to explain screening decisions clearly
Fluent English proficiency

Who Should Apply

Apply if you regularly use a code-agent CLI and can produce trustworthy records of real engineering work from start to finish. This opportunity is especially suited to professional software engineers who are comfortable moving through unfamiliar repositories, investigating bugs, implementing changes, refactoring code, running tests, and reviewing final diffs.

The best contributors will combine strong technical judgment with disciplined documentation. You should be able to distinguish successful and unsuccessful outcomes, explain why a trace meets the requirements, and submit complete evidence without missing important context.

Expert software engineers with hands-on development experience
Engineers comfortable working across multiple programming stacks
Contributors who can validate results through tests and repository evidence
Professionals interested in helping train the next generation of AI coding agents

How to Apply Through OpenTrain

Create a free OpenTrain account and apply for this contractor project in minutes. Be ready to discuss your relevant software engineering and AI coding agent experience, your daily code-agent CLI usage, and your ability to produce complete, validated execution traces.

If selected, you will receive the applicable project workflow and submission expectations, including trace formats, complexity tiers, delivery targets, and verification standards.

Create a free OpenTrain account
Submit your application and relevant experience
Complete the required screening discussion
Follow the project requirements for trace creation and validation
Submit verified traces for per-trace payment

Real Claude Code CLI Execution Traces (Code Agents)

Share this project