Back to jobs

STEM Jobs in India

Rex.zone
United States
Full-time
3,000 – 5,000 / year
AI tools:
LLM
Applications go directly to the hiring team

Full Description

STEM Jobs in India (Remote, Full-Time)

Rex.zone connects candidates to engineering and applied AI/ML workflows that power large language model training pipelines. You will support training data quality through data labeling, RLHF evaluation, QA evaluation, and guideline-based reviews across NLP, computer vision, and content safety domains.

What You Will Do

* Execute data labeling for NLP and computer vision using clear taxonomies and instructions

* Perform RLHF-style preference ranking and rubric-based scoring for LLM outputs

* Run QA evaluation for completeness, consistency, and annotation guidelines compliance

* Conduct prompt evaluation and error categorization to surface model failure modes

* Perform named entity recognition (NER) and span labeling with strict boundary rules

* Support content safety labeling (e.g., harassment, self-harm, regulated goods)

* Maintain audit trails, adjudication notes, and participate in calibration sessions

* Communicate findings to stakeholders to drive measurable model performance improvement

Required Qualifications

* STEM background or equivalent engineering experience

* High attention to detail; ability to follow detailed specifications and handle edge cases

* Experience with QA evaluation, data labeling, or structured review workflows

* Strong documentation skills and consistency at scale in a metrics-driven environment

Preferred Qualifications

* Experience with RLHF, LLM evaluation, prompt evaluation, or ranking tasks

* Familiarity with NER, text classification, or computer vision annotation

* Exposure to content safety labeling and policy categories

* Experience improving training data quality via audits and inter-annotator agreement

Tools & Workflows

Web-based annotation tools, spreadsheet QA workflows, issue tracking, versioned guidelines, and structured rubrics.

Work is designed for distributed collaboration.

Applications go to the hiring team directly