AI Training Specialist (Brazil, Remote)
6 - 10 years
United States
Full-time
Remote
1 month ago
About the role
AI Training Jobs in Brazil
Rex.zone connects Brazil-aligned talent with remote, full-time AI training work supporting LLM and computer vision model improvement through high-quality data and evaluation signals.
About The Role
You will contribute to AI training workflows that turn raw text, images, and model outputs into supervised datasets and evaluation signals. Work includes data labeling, RLHF preference ranking, prompt evaluation, QA evaluation, and guideline-based review in asynchronous, quality-driven delivery.
What You’ll Do
Domains You May Support
NLP/LLM evaluation, content safety labeling, multilingual Portuguese/English tasks, and computer vision annotation (e.g., bounding boxes/segmentation), depending on project needs.
Requirements
Compensation
Hourly range: $30–$50/hr (USD).
Quality is measured by audit accuracy, guidelines compliance, inter-annotator agreement, defect rates, and turnaround time.
Rex.zone connects Brazil-aligned talent with remote, full-time AI training work supporting LLM and computer vision model improvement through high-quality data and evaluation signals.
About The Role
You will contribute to AI training workflows that turn raw text, images, and model outputs into supervised datasets and evaluation signals. Work includes data labeling, RLHF preference ranking, prompt evaluation, QA evaluation, and guideline-based review in asynchronous, quality-driven delivery.
What You’ll Do
- Create and apply annotation taxonomies and guidelines
- Label text and image data; perform named entity recognition and document classification
- Conduct RLHF comparisons and preference ranking
- Execute prompt evaluation and rubric-based scoring
- Complete QA evaluation, adjudication, and error analysis
- Validate training data quality via sampling and audits
- Track inter-annotator agreement, defect rates, and calibration outcomes
- Document decisions to support consistent evaluation and dataset versioning
Domains You May Support
NLP/LLM evaluation, content safety labeling, multilingual Portuguese/English tasks, and computer vision annotation (e.g., bounding boxes/segmentation), depending on project needs.
Requirements
- Mid-senior capability in structured evaluation and operational rigor
- Experience with data labeling or QA evaluation in production settings
- Comfort with ambiguity resolution and rubric adherence
- Strong written communication for guideline interpretation
- Familiarity with LLM evaluation concepts (helpfulness, factuality, toxicity, instruction following)
Compensation
Hourly range: $30–$50/hr (USD).
Quality is measured by audit accuracy, guidelines compliance, inter-annotator agreement, defect rates, and turnaround time.
Skills
Human ResourcesTechnology, Information and Internet