STEM Careers in Brazil

About The Role

STEM Careers in Brazil at Rex.zone support AI/ML training workflows and large language model (LLM) evaluation. In this full-time remote engineering role, you will help improve model performance through high-quality training data, RLHF evaluation, and rigorous QA processes.

What You Will Do

* Support LLM training pipelines by validating datasets, rubrics, and evaluation protocols used in RLHF and offline evaluation

* Perform data labeling and QA evaluation across NLP, named entity recognition (NER), and computer vision annotation tasks

* Execute prompt evaluation and response grading to improve helpfulness, correctness, and policy compliance

* Apply annotation guidelines compliance checks, identify inconsistencies, and propose guideline clarifications

* Track training data quality metrics (agreement rates, defect taxonomy, escalation patterns) and recommend fixes

* Contribute to content safety labeling workflows (toxicity, self-harm, hate/harassment, privacy, and sensitive content categories)

* Collaborate asynchronously with cross-functional stakeholders (engineering, ops, QA) to unblock delivery and improve throughput

Required Qualifications

* Mid-Senior experience in a STEM discipline (engineering, computer science, data, or related technical field) or equivalent practical experience

* Hands-on experience with at least two of: data labeling, QA evaluation, prompt evaluation, RLHF workflows, or dataset auditing

* Strong written communication for producing clear guidelines, rubrics, and defect reports

* Working familiarity with NLP concepts (classification, NER, information extraction) and/or computer vision annotation

* Ability to follow strict annotation guidelines compliance while applying sound judgment to ambiguous edge cases

* Experience collaborating remotely across time zones with consistent delivery in full-time schedules

FAQ

Is this role remote?

Yes. Remote Type is Remote and the role is designed for full-time remote work.

Why does a STEM engineering role include RLHF, data labeling, and LLM evaluation?

Modern AI systems depend on high-quality training data and robust evaluation. This role supports LLM training pipelines through RLHF datasets, prompt evaluation, QA evaluation, and annotation guidelines compliance to drive model performance improvement.