For employers

Hire this AI Trainer

Sign in or create an account to invite AI Trainers to your job.

LoginCreate account
Invite to Job
J
Jili Xu

Jili Xu

AI Data Annotator——healthcare

China flagShenzhen, China
$20.00/hrIntermediateDon T Disclose

Key Skills

Software

Don't disclose

Top Subject Matter

Medical image analysis
pancreatic cancer immunogenomics
Medical image segmentation

Top Data Types

ImageImage
TextText
Computer Code ProgrammingComputer Code Programming

Top Task Types

ClassificationClassification
SegmentationSegmentation
Question AnsweringQuestion Answering
Text SummarizationText Summarization
Text GenerationText Generation
Computer Programming/CodingComputer Programming/Coding
Point/Key PointPoint/Key Point

Freelancer Overview

My experience spans multiple machine learning and data-intensive research projects in biomedical and clinical settings, where data preparation and quality were foundational to every outcome. At Hong Kong Polytechnic University, I designed end-to-end data analysis pipelines integrating statistical methods and machine learning models, applying algorithms such as logistic regression, random forest, and XGBoost to multi-omic datasets. I also optimized deep learning segmentation models (nnU-Net) with customized preprocessing, augmentation, and post-processing pipelines — work that required meticulous attention to data quality, annotation consistency, and ground-truth validation against metrics like Dice score and Hausdorff distance. Beyond pipeline work, I have hands-on experience with complex, multi-modal datasets — including histopathological slides, genomic data, and clinical records — where feature selection, data cleaning, and interpretability are critical. My independent project applying Multiple Instance Learning (MIL) to pancreatic cancer histology required curating and structuring weakly labeled image data to train and evaluate four different MIL architectures. Combined with proficiency in Python (PyTorch, Sklearn, Pandas, NumPy) and R, and a strong research publication record, I bring both the technical rigor and domain expertise to contribute meaningfully to AI training data and labeling workflows.

IntermediateEnglish

Labeling Experience

Medical Image Segmentation Labeling - Research Assistant, HK Polytechnic University

Segmentation
I optimized nnU-Net architectures to improve automated segmentation of organs in medical images for radiotherapy planning. The work included preprocessing medical imaging datasets, custom augmentations, and benchmarking segmentation accuracy of key prostate cancer structures. This project improved AI model performance for automated treatment planning. • Prepared DICOM datasets for segmentation model training and evaluation • Labeled segmentations for urethra, bladder, rectum, and penile bulb • Benchmarked results using Dice score and Hausdorff distance • Documented segmentation protocols for model development

I optimized nnU-Net architectures to improve automated segmentation of organs in medical images for radiotherapy planning. The work included preprocessing medical imaging datasets, custom augmentations, and benchmarking segmentation accuracy of key prostate cancer structures. This project improved AI model performance for automated treatment planning. • Prepared DICOM datasets for segmentation model training and evaluation • Labeled segmentations for urethra, bladder, rectum, and penile bulb • Benchmarked results using Dice score and Hausdorff distance • Documented segmentation protocols for model development

2025 - Present

Histopathology Image Labeling for Gene Signature Identification - Independent Project

Classification
The project involved identifying gene signatures from pancreatic cancer histopathological images using Multiple Instance Learning (MIL). I prepared and processed medical image data for training and evaluation of MIL models and selected optimal models for gene activation prediction. The work contributed to improving immunotherapeutic strategies for pancreatic cancer. • Labeled and annotated large sets of histology slide images • Utilized MIL methods to generate and verify gene signature predictions • Evaluated MIL model performance for classification accuracy • Documented annotation protocols and standards used

The project involved identifying gene signatures from pancreatic cancer histopathological images using Multiple Instance Learning (MIL). I prepared and processed medical image data for training and evaluation of MIL models and selected optimal models for gene activation prediction. The work contributed to improving immunotherapeutic strategies for pancreatic cancer. • Labeled and annotated large sets of histology slide images • Utilized MIL methods to generate and verify gene signature predictions • Evaluated MIL model performance for classification accuracy • Documented annotation protocols and standards used

2023 - 2023

Education

G

Georgia Institute of Technology

Master, Biomedical Engineering

Master
2026 - 2026
F

Fudan University

Doctor of Philosophy, Biological and Pharmaceutical Engineering

Doctor of Philosophy
2022 - 2023

Work History

T

The First Affiliated Hospital Of Zhejiang Chinese Medical University

Resident Doctor

Hangzhou
2019 - 2022