Liu I.

Key Skills

Software

LabelImg

Label Studio

Top Subject Matter

Customer Service AI

Intent Recognition

Government Hotline

Top Data Types

Text

Audio

Image

Top Task Types

Entity (NER) Classification

Emotion Recognition

Freelancer Overview

AI Trainer at Zhilian Technology. Core strengths include LabelImg and Label Studio. Education includes Bachelor of Science, The University of Hong Kong (2023). AI-training focus includes data types such as Text and Audio and labeling workflows including Entity (NER) Classification and Emotion Recognition.

Labeling Experience

AI Trainer at Zhilian Technology

LabelImgTextEntity (NER) Classification

Led the intelligent customer service system corpus optimization for intent recognition. Developed annotation specifications for core intent categories, and implemented a three-level quality control framework. Trained annotation team and applied active learning to enhance annotation precision. • Improved intent recognition accuracy from 78% to 92% in three months • Established comprehensive annotation guidelines for 12 core intents • Used LabelImg and Label Studio for standardized labeling • Iteratively updated annotation protocols to maintain over 90% system accuracy

2024 - Present

Project Lead – Cantonese Speech Annotation

Label StudioAudioEmotion Recognition

Managed end-to-end corpus creation and annotation for a government AI voice assistant. Developed and deployed a triple annotation system emphasizing emotion, intent, and entity recognition. Delivered a quality-controlled, multi-category speech dataset for production use. • Covered 15 categories of livelihood inquiries with over 8,000 data points • Set company annotation standard with versioned specification • Enhanced model ability to detect complaints and complex user intent • Optimized speech annotation process to reduce QA burden

2023 - 2024

Education