Zijun F.

Key Skills

Software

Prodigy

Other

Top Subject Matter

Bilingual (English/Chinese) AI evaluation and structured data QA for sports-science and prompt-following contexts.

Image labeling for movement/posture and sports-science related structured datasets (English/Chinese review context).

Sports-science experimental data collection and QA (movement-video analysis and sensor-based testing workflows).

Top Data Types

Text

Image

Video

Document

Top Task Types

Classification

Data Collection

Freelancer Overview

Independent AI Workflow Builder / Data Annotation Projects (AI evaluation, fact-checking, rubric scoring). Brings 6+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Prodigy and Other. Education includes Master of Science, Beijing Sport University (2025) and Bachelor of Science, Beijing Sport University (2025). AI-training focus includes data types such as Text, Image, and Video and labeling workflows including Evaluation, Rating, and Classification.

Labeling Experience

Independent AI Workflow Builder / Data Annotation Projects | Beijing, China

Image

No description provided.

2025 - Present

Independent Workflow Builder (AI Evaluation/Data QA) - Independent AI Workflow Builder

Image

Built and operated an independent data-annotation workflow for AI evaluation and model-training preparation. Led structured review and quality-control steps to verify factuality, clarity, and instruction compliance in generated outputs. Applied bilingual prompt iteration and rubric-based scoring to improve reliability and usability of the resulting labeled datasets. • Used Prodigy for structured annotation workflow design including review, quality control, and export steps • Performed source-grounded fact-checking to flag unsupported claims, inconsistencies, and missing context • Conducted image annotation for target regions, postures, categories, and visually relevant features • Applied accept/reject/ignore decisions, rubric-based scoring, and prompt optimization using LLM tools

2025 - Present

Independent AI Workflow Builder / Data Annotation Projects (image annotation and dataset QA)

ProdigyImageClassification

Created structured image annotations by labeling target regions, movement postures, object categories, and visually relevant features for image-based datasets. Applied accept/reject/ignore review decisions to ensure quality and rubric-aligned annotation usefulness. Cleaned datasets by detecting duplicates, inconsistent labels, formatting errors, and incomplete or low-quality samples prior to export. • Labeled target regions and visually salient features for structured datasets. • Applied label consistency and completeness review to reduce noise. • Performed anomaly and low-quality sample checks. • Exported labeled outputs in structured formats (e.g., JSONL/CSV) for downstream tasks.

2025 - Present

Independent AI Workflow Builder / Data Annotation Projects (AI evaluation, fact-checking, rubric scoring)

ProdigyText

Provided rubric-based evaluation for model outputs, including factual accuracy, clarity, instruction-following, tone, consistency, and usefulness. Performed source-grounded fact-checking by comparing AI outputs against original materials and flagging unsupported claims, inconsistencies, missing context, or misleading wording. Applied accept/reject/ignore decisions to guide model-output QA and improvement iterations. • Wrote and optimized prompts for evaluation tasks using ChatGPT and Codex. • Conducted hallucination detection via source-response mismatch checks. • Reviewed responses for clarity, tone, and guideline compliance. • Rewrote/adjusted outputs based on evaluation findings.

2025 - Present

Sports Science Research Data Assistant - Beijing Sport University

Video

Served as a sports science research data assistant supporting the collection, cleaning, and organization of experimental datasets. Worked with specialized lab and testing equipment to transform raw measurements and video-derived information into structured tables and summaries. Maintained rigorous data consistency and QA procedures to ensure reliable analysis-ready outputs across repeated testing rounds. • Collected and organized experimental datasets for 100+ participants across 3 testing rounds with ~1,000 entries • Managed raw data in Excel by checking for missing or inconsistent records and standardizing formats • Supported performance evaluation using iEMG, force plate testing, CMJ, RSI, RFD, braking impulse, and movement-video analysis • Ensured standardized labeling, anomaly checks, and repeatable QA processes in controlled research environments

2021 - Present

Show more

Education

Work History