F W.

Key Skills

Software

CVAT

Top Subject Matter

Generative AI image evaluation and preference learning (RLHF) for image generation quality.

Computer vision quality assessment for generative image models.

Multimedia Content Generation & Evaluation

Top Data Types

Image

Text

Document

Top Task Types

RLHF

Freelancer Overview

AI Product & Data Training Specialist with a Master's degree from a top-tier Chinese university (211 & Double First-Class). Over 20 years of professional experience in software development, internet product operations, and technology team management. Since 2023, fully dedicated to AI model training and evaluation, with deep hands-on expertise in RLHF (preference ranking and response evaluation), generative image quality assessment, and prompt engineering. Led the design and implementation of recursive testing frameworks to improve the reliability and performance of generative AI outputs. Managed large-scale image dataset ranking and annotation workflows focused on output validation and quality enhancement. Developed an end-to-end AI product prototype covering data collection, model inference, and performance iteration. Proven ability to combine technical rigor with creative evaluation, delivering measurable improvements in model aesthetics and prompt adherence. Native Chinese speaker with fluent English, ready for high-complexity AI training tasks across image, text, and multimodal domains.

Labeling Experience

AI Product & Data Training Specialist (Self-Directed R&D) - Self-Directed R&D

ImageRLHF

Designed and implemented recursive testing frameworks to improve the quality and reliability of generative AI outputs. Led an AI workflow project focused on AI-generated wallpaper and content creation for a major social platform. Performed large-scale evaluation and ranking of image datasets to enhance aesthetics, prompt adherence, and overall model performance. • Built recursive testing/RLHF-style evaluation logic for generative models • Managed dataset ranking and annotation workflows focused on output validation, not labeling • Developed an end-to-end AI product prototype from data collection to inference • Applied prompt engineering and image generation model evaluation techniques

2023 - Present

Model Output Evaluation & Dataset Ranking/Annotation (within AI R&D)

Don't discloseImageClassification

Conducted large-scale evaluation and quality assessment of generative model outputs. Produced labeled judgments used to refine image generation performance and model evaluation criteria. Focused on improving the alignment between generated images and intended prompts. • Evaluated and rated image outputs for quality. • Captured label signals supporting model performance tuning. • Used evaluation results to adjust prompts and iteration strategy. • Supported dataset update/selection for better output consistency.

2023 - Present

AI Product & Data Training Specialist (Self-directed R&D)

Don't discloseImageRLHF

Performed RLHF-style preference ranking and recursive evaluation to improve generative image outputs. Labeled and ranked image datasets based on aesthetic quality and prompt adherence to guide model behavior. Iterated evaluation and annotation cycles to achieve higher market-ready output quality. • Preference ranking of model responses for generative images. • Image dataset annotation aligned with aesthetics and prompt constraints. • Recursive testing/evaluation loops (RLHF workflow logic). • Quality assessment to drive prompt adherence improvements.

2023 - Present

Education

Work History