Moses A.

Key Skills

Software

Remotasks

Scale AI

Telus

Top Subject Matter

Artificial Intelligence – RLHF, LLM Evaluation & Prompt Engineering

Computer Vision – Image Annotation, Visual Grounding & Image Evaluation

Technology – Data Annotation, Quality Assurance & AI Model Training

Top Data Types

Image

Video

Text

Top Task Types

Bounding Box

Object Detection

Text Generation

RLHF

Text Summarization

Evaluation/Rating

Data Collection

Prompt + Response Writing (SFT)

Freelancer Overview

I have over two years of experience in AI data annotation, evaluation, and training data quality assurance, working across platforms such as Outlier, Multimango, Stellar, and Joinstellar. My work has involved evaluating AI-generated text and image outputs for accuracy, instruction adherence, logical consistency, and overall quality to support reinforcement learning from human feedback (RLHF) and model improvement. I have extensive experience conducting ELO-based image comparisons, visual grounding quality assurance, reference-to-image (R2I) assessments, image captioning, prompt evaluation, entity tagging, and data categorization. I am highly skilled at identifying model weaknesses, edge cases, and failure patterns while maintaining strict adherence to annotation guidelines and quality rubrics. What sets me apart is my combination of strong analytical thinking, attention to detail, and technical background in both engineering and web development. I consistently deliver high-quality annotations with low revision rates and fast turnaround times. My experience includes crafting detailed image captions, evaluating complex multimodal AI outputs, creating adversarial visual reasoning tasks, and providing structured feedback that contributes to prompt refinement and dataset improvement. Combined with strong communication skills and a disciplined, process-driven approach, I am well-equipped to contribute to AI training, data labeling, prompt engineering, and model evaluation projects.

Labeling Experience

Multimodal Image Evaluation & Visual Grounding Specialist

ImageText Generation

Contributed to AI training initiatives focused on computer vision and generative image models by evaluating thousands of AI-generated images using structured quality assessment frameworks. Performed ELO-based ranking of image outputs, assessing factors such as prompt adherence, visual realism, object placement, composition, lighting consistency, and style accuracy. Evaluated image editing tasks including background replacement, object removal, style transfer, and content preservation to improve model performance and reliability. Performed visual grounding and spatial reasoning quality assurance by creating and validating complex image-based queries designed to test model understanding of object relationships, locations, and scene context. Produced detailed image captions and metadata annotations describing visual content, attributes, and interactions to support training dataset development. Consistently maintained high annotation accuracy and quality standards while identifying edge cases and model failure patterns that informed model refinement and training improvements.

2025 - Present

AI Data Annotator & RLHF Evaluator

VideoBounding Box

Worked as an AI Data Annotator and Evaluator on large-scale AI training and reinforcement learning (RLHF) projects through Outlier, Multimango, and Stellar. Evaluated AI-generated text and image outputs for accuracy, instruction adherence, logical consistency, safety, and overall quality. Conducted ELO-based pairwise ranking of text-to-image generations, assessed image editing tasks involving object removal, style transfer, and identity preservation, and performed multi-reference image evaluation (R2I) to measure prompt alignment and compositional accuracy.

2023 - 2026

RLHF Data Trainer & AI Response Quality Analyst

TextRLHF

Supported the training and optimization of large language models through Reinforcement Learning from Human Feedback (RLHF) workflows. Evaluated AI-generated responses across diverse domains including STEM, coding, research, creative writing, and general knowledge. Assessed outputs for factual accuracy, reasoning quality, instruction adherence, clarity, tone, safety, and overall usefulness using detailed evaluation guidelines and scoring frameworks. Performed comparative ranking and preference selection tasks to identify higher-quality model responses, contributing directly to model alignment and performance improvements. Annotated edge cases, documented reasoning errors, and provided structured feedback to improve training datasets and evaluation standards. Demonstrated strong attention to detail, consistency, and analytical judgment while maintaining high productivity and quality across large-scale annotation projects.

2023 - 2025

Education

Work History