The work
You will combine full-stack development with hands-on AI model training and evaluation. The work includes building reliable code, reviewing model responses, creating training data, and explaining your decisions clearly.
You will work with researchers and annotators to improve model quality across coding and other technical tasks.
- Design, develop, and maintain readable, reusable, secure JavaScript or TypeScript code.
- Evaluate model responses against defined criteria, rank answers, benchmark performance, and analyze results.
- Recommend improvements and write clear explanations for evaluation decisions.
- Create and maintain datasets for supervised fine-tuning.
- Support reinforcement learning with human feedback and reward-model refinement.
- Write and improve model responses for clarity, relevance, and technical accuracy.
- Review code and documentation, then provide constructive feedback.
What it pays and takes
The role details do not list a pay rate. This is a remote contractor assignment with a one-month contract duration.
- Hours: At least four hours per day and 20 hours per week, with options for 20, 30, or 40 hours weekly.
- Schedule: Four hours of overlap with Pacific Time is required.
- Location: Open worldwide and fully remote.
- Language: Excellent spoken and written English.
- JavaScript or TypeScript: Extensive experience and a strong command of syntax and conventions.
- Technical skills: Strong knowledge of ES6 and at least one of Node.js, Nest.js, React, Angular, or Vue.
- Development: Experience building modular web applications with scalable architectures.
- Code quality: Ability to write organized, maintainable, correct, and clearly annotated code.
- Education: An engineering or computer science degree, or equivalent experience.
- Valuable additional experience: Docker and software quality assurance or test planning.
How it works
Apply on OpenTrain with your resume, then complete the application on the hiring site.
About AI training work
AI training is the human work behind modern artificial intelligence, including writing and rating model responses, evaluating code, and preparing examples for training. People with strong technical skills help models produce more accurate, useful, and reliable results.