Yu H.

Key Skills

Software

Mighty AI

Top Subject Matter

AI Data Annotation & Governance

Computer Vision Data Preparation

NLP Training Data Processing

Top Data Types

Image

Top Task Types

Bounding Box

Freelancer Overview

I possess solid AI professional literacy from South China University of Technology and rich full-cycle practical experience in AI training data processing. I am skilled in standardized data annotation, quality inspection, data cleaning and sample optimization for computer vision and NLP scenarios, proficient in mainstream annotation tools and Python data processing technology, and have stably maintained a labeling accuracy rate of over 98% in large-scale dataset production. With an in-depth understanding of deep learning model training logic, I can dynamically optimize annotation rules and dataset structure based on model feedback, solve problems such as sample imbalance and data noise, and realize data-model collaborative optimization. I have supported data preparation for multiple industrial vision and large model RAG projects, capable of efficiently outputting high-quality training data to drive model iteration and performance improvement.

Labeling Experience

AI Algorithm R&D Engineer, Data Annotation Lead

ImageBounding Box

Led and managed the entire process of data annotation for computer vision and NLP datasets, including large-scale image object detection and text entity annotation. Established standardized operation guidelines and implemented a rigorous quality inspection system to ensure high annotation accuracy and consistency. Collaborated with R&D teams to iteratively optimize annotation rules based on model training feedback and continuously improved dataset quality for deep learning models. • Conducted fine annotation, audit, and correction of hundreds of thousands of high-definition images and industry texts for object detection, defect classification, and intent recognition. • Built and improved internal annotation pipelines and semi-automatic annotation processes, achieving a 40% boost in data preparation efficiency. • Administered quality verification workflows using a multi-level inspection approach, effectively sustaining annotation accuracy above 98%. • Provided crucial data support for over ten AI projects, driving model recognition accuracy gains from 76% to 93%.

2022 - 2024

AI R&D Intern, Data Annotation & Governance

ImageBounding Box

Assisted in multi-scenario data annotation and preprocessing for intelligent image detection projects. Performed fine-grained labeling (object framing, defect classification, point marking) and adhered to strict project annotation standards on tens of thousands of images. Collaborated on correction, verification, archiving, and dataset structuring for optimal AI model training. • Supported project teams in precise annotation, screening, and correction of non-standard or erroneous samples. • Standardized dataset splits (training, validation, test) and maintained data traceability through meticulous archiving. • Documented labeling logs and summarized frequent annotation problems for iterative dataset improvements. • Helped deliver industrial product datasets that enabled accurate model development and deployment.

2020 - 2020

Education

Work History