R

STEM Careers India Engineer

Accepting applications

Rex.zone · United States

Full-Time Mid_senior AIaiaterf
Posted
2 May
Category
Test
Experience
Mid_senior
Country
United States
About Rex.zone

Rex.zone connects India-aligned STEM talent with Remote, Full-Time engineering programs supporting AI/ML training workflows. You will help improve training data quality and evaluation rigor across LLM training pipelines, RLHF, and model evaluation.

About The Role

You will build and operate scalable data labeling and evaluation workflows, implement annotation tooling, run prompt evaluation, and partner with distributed teams to drive measurable model performance improvement.

Key Responsibilities

Design and improve LLM training pipelines for supervised fine-tuning and RLHF
Build and maintain annotation tooling, task routing, and QA evaluation checks
Define labeling taxonomies for NLP tasks (e.g., named entity recognition, classification)
Support computer vision annotation workflows (bounding boxes, polygons, segmentation QA)
Run prompt evaluation and rubric-based model evaluation to identify failure modes
Implement content safety labeling policies, audits, and escalation paths
Create annotation guidelines, sampling plans, and ensure annotation guidelines compliance
Analyze training data quality metrics, disagreement patterns, and error distributions
Document datasets and evaluation artifacts for traceability and reproducibility

Required Qualifications

Mid-Senior experience in engineering, data operations engineering, or applied ML workflows
Hands-on experience with data labeling systems, QA evaluation, or model evaluation pipelines
Familiarity with RLHF concepts and large language model evaluation methods
Working knowledge of NLP and/or computer vision annotation
Strong written communication for remote collaboration

How To Apply

Apply via Rex.zone and highlight experience in training data quality, annotation guidelines compliance, QA evaluation, RLHF, prompt evaluation, NLP/NER, computer vision annotation, and content safety labeling.
Show more Show less