Jianhui L.

Key Skills

Software

Other

Top Subject Matter

Financial Compliance & Risk Analysis

Medicine Domain Expertise

Environmental Monitoring

Top Data Types

Text

Image

Top Task Types

Question Answering

Segmentation

Freelancer Overview

Retrieval-Augmented Generation (RAG) for Domain QA. Brings 1+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other. Education includes Bachelor of Commerce and Bachelor of Computer Science, University of New South Wales (2025). AI-training focus includes data types such as Text and Image and labeling workflows including Question Answering and Segmentation.

Labeling Experience

Aerial Imagery for Dead Tree Segmentation (Kaggle)

OtherImageSegmentation

This Kaggle challenge required semantic segmentation of dead trees using high-resolution aerial imagery. The participant annotated image data to train and refine AI segmentation models, using advanced deep learning architectures. The process showcased the labeling and evaluation of meaningful geospatial features for environmental monitoring.• Contributed to segmentation of trees in geospatial imagery for AI training.• Selected and annotated data samples for modeling UNet++ and SegFormer architectures.• Employed semantic segmentation label types on remote sensing imagery datasets.• Validated annotation quality by evaluating competition performance metrics.

2024 - 2024

Retrieval-Augmented Generation (RAG) for Domain QA

OtherTextQuestion Answering

This project involved building a domain knowledge base using ScienceQA datasets and evaluating retrieval-augmented generation for domain-specific question answering. The process included organizing datasets in finance and medicine and using AI models to answer domain questions based on this data. GPT-4o's performance on seen questions improved significantly through these experiments.• Built QA datasets in finance and medical domains using ScienceQA.• Set up retrieval-augmented generation (RAG) pipelines with LangChain and external VectorDB.• Evaluated GPT-4o and Claude on in-domain question answering tasks.• Demonstrated significant accuracy gains for domain-specific LLM assistants.

2024 - 2024

Education

Work History