Tsai Y

Key Skills

Software

Remotasks

Scale AI

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Image

Text

Top Task Types

Evaluation Rating

Fine Tuning

Prompt Response Writing SFT

Text Generation

Translation Localization

Freelancer Overview

Highly proficient in Large Language Model (LLM) Content Evaluation and Training Data Review. My consistent high quality led to rapid promotion to a Senior Reviewer/Quality Assurance (QA) role, where I manage and optimize content from other contributors to ensure data integrity and consistency. I specialize in evaluating content for accuracy, logic, and adherence to complex guidelines, providing specific, actionable feedback for model enhancement. Leveraging a professional background in digital marketing and UI/UX, I bring a unique, detail-oriented and user-centric focus to content structure. I am highly proficient in diverse content training areas, including Travel and Itinerary Planning, Creative Rewriting & Linguistic Refinement, and complex Data Organization & Summarization. My background ensures that the training data I handle is not only technically correct but also contextually relevant and engaging for end-users.

Labeling Experience

Audio-to-Text LLM Response Evaluation & Usability Testing

Scale AIAudioClassificationTranslation Localization

Scope: Evaluated LLM performance following voice input on mobile applications. Tasks included: Speech-to-Text Transcription (Mapping user queries), Response Annotation (Pasting the model's text output), and Quality Rating based on predefined rubrics. This project tested model comprehension under voice command constraints.

Scope: Evaluated LLM performance following voice input on mobile applications. Tasks included: Speech-to-Text Transcription (Mapping user queries), Response Annotation (Pasting the model's text output), and Quality Rating based on predefined rubrics. This project tested model comprehension under voice command constraints.

2024 - 2025

LLM Training Task at Outlier (Powered by Scale AI)

Scale AITextClassificationQuestion Answering

Scope: Continuous data contribution for LLMs, focused on alignment, accuracy, and safety. Covered creation, evaluation, and QA. Specific Tasks & Advanced Skills: Content Competition (LLM vs. Human): Authored superior "gold standard" content for model training, leveraging Creative Rewriting and Logical Structuring. Error Correction (Red Teaming): Specialized in identifying logical flaws, factual errors, and crafting corrective instruction/response pairs to enhance model robustness. Data Optimization (QA/Reviewer): Advanced to a designated Reviewer role. Performed multi-layer review of peer data, serving as a critical quality gatekeeper for data integrity. Quality Measures & Focus: Evaluation centered on factual accuracy, logical coherence, and user intent/helpfulness. Meticulous approach (informed by UI/UX) ensured data met complex annotation rubrics.

Scope: Continuous data contribution for LLMs, focused on alignment, accuracy, and safety. Covered creation, evaluation, and QA. Specific Tasks & Advanced Skills: Content Competition (LLM vs. Human): Authored superior "gold standard" content for model training, leveraging Creative Rewriting and Logical Structuring. Error Correction (Red Teaming): Specialized in identifying logical flaws, factual errors, and crafting corrective instruction/response pairs to enhance model robustness. Data Optimization (QA/Reviewer): Advanced to a designated Reviewer role. Performed multi-layer review of peer data, serving as a critical quality gatekeeper for data integrity. Quality Measures & Focus: Evaluation centered on factual accuracy, logical coherence, and user intent/helpfulness. Meticulous approach (informed by UI/UX) ensured data met complex annotation rubrics.

2024 - 2025

Education

Work History