Rebecca Zhao

Key Skills

Software

Labelbox

Top Subject Matter

No subject matter listed

Top Data Types

Image

Text

Video

Top Label Types

Bounding Box

Point Key Point

Segmentation

Classification

Tracking

Text Summarization

RLHF

Fine Tuning

Evaluation Rating

Prompt Response Writing SFT

Action Recognition

Freelancer Overview

I am an AI training specialist with 3 years of hands-on experience in data labeling and annotation for machine learning and computer vision projects. My expertise includes annotating large-scale image, video, and audio datasets using tools like Labelbox, CVAT, Supervisely, Roboflow, and Amazon SageMaker Ground Truth. I am skilled in object detection, semantic segmentation, multi-object tracking, and audio transcription, consistently maintaining high accuracy and data quality. I have worked closely with machine learning engineers to refine labeling guidelines, optimize workflows, and ensure datasets are YOLO-ready for model training. Additionally, I have experience training and onboarding new annotators, and I am passionate about delivering precise, high-quality datasets that drive AI model performance.

Labeling Experience

Smart Surveillance Video Annotation & Multi-Object Tracking

LabelboxVideoBounding BoxSegmentation

Currently contributing to a smart surveillance AI project focused on training computer vision models for real-time object detection and behavioral analysis. The project involves annotating large-scale CCTV and drone-captured video datasets for security monitoring applications. To date, over 45,000+ video frames have been accurately labeled across multiple urban and commercial environments. Key responsibilities include: Drawing bounding boxes around people, vehicles, and security-relevant objects. Performing multi-object tracking while maintaining consistent object IDs across sequential frames. Annotating human activities for action recognition (e.g., running, loitering, entering restricted areas). Applying segmentation masks for high-precision object boundaries when required. Reviewing and validating annotations for spatial accuracy and temporal consistency. Exporting datasets in YOLO and COCO formats for model training and evaluation. Quality assurance measures followed inclu

Currently contributing to a smart surveillance AI project focused on training computer vision models for real-time object detection and behavioral analysis. The project involves annotating large-scale CCTV and drone-captured video datasets for security monitoring applications. To date, over 45,000+ video frames have been accurately labeled across multiple urban and commercial environments. Key responsibilities include: Drawing bounding boxes around people, vehicles, and security-relevant objects. Performing multi-object tracking while maintaining consistent object IDs across sequential frames. Annotating human activities for action recognition (e.g., running, loitering, entering restricted areas). Applying segmentation masks for high-precision object boundaries when required. Reviewing and validating annotations for spatial accuracy and temporal consistency. Exporting datasets in YOLO and COCO formats for model training and evaluation. Quality assurance measures followed inclu

2023

Autonomous Vehicle Video Annotation & Multi-Object Tracking Project

LabelboxImageBounding BoxPoint Key Point

Worked on a large-scale autonomous vehicle computer vision project involving video annotation and object detection dataset preparation. The project included annotating over 50,000+ video frames captured from urban and highway driving environments.

Worked on a large-scale autonomous vehicle computer vision project involving video annotation and object detection dataset preparation. The project included annotating over 50,000+ video frames captured from urban and highway driving environments.

2023 - 2025

LLM Fine-Tuning & RLHF Data Annotation Project

LabelboxTextText SummarizationRLHF

Contributed to a large-scale Large Language Model (LLM) training and fine-tuning project focused on improving response accuracy, safety, and contextual understanding. Key responsibilities included: Writing high-quality prompt + response pairs for supervised fine-tuning (SFT). Ranking multiple AI-generated responses based on relevance, helpfulness, and safety (RLHF process). Performing detailed evaluation and rating of model outputs using structured rubrics. Annotating text datasets for summarization and classification tasks. Identifying hallucinations, bias, and policy violations in generated outputs. Providing structured feedback to improve model alignment and performance. The project involved reviewing and annotating over 25,000+ prompts and responses across multiple domains including technology, healthcare, education, and general knowledge.

Contributed to a large-scale Large Language Model (LLM) training and fine-tuning project focused on improving response accuracy, safety, and contextual understanding. Key responsibilities included: Writing high-quality prompt + response pairs for supervised fine-tuning (SFT). Ranking multiple AI-generated responses based on relevance, helpfulness, and safety (RLHF process). Performing detailed evaluation and rating of model outputs using structured rubrics. Annotating text datasets for summarization and classification tasks. Identifying hallucinations, bias, and policy violations in generated outputs. Providing structured feedback to improve model alignment and performance. The project involved reviewing and annotating over 25,000+ prompts and responses across multiple domains including technology, healthcare, education, and general knowledge.

2023 - 2023

Education

Work History