For employers

Hire this AI Trainer

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

LoginCreate account
Invite to Job
Xizi Yu

Xizi Yu

LLM Evaluation and Text Generation Specialist in English & Chinese

Canada flagVictoria, Canada
$15.00/hrEntry LevelInternal Proprietary Tooling

Key Skills

Software

Internal/Proprietary Tooling

Top Subject Matter

Music Recording
Translation/Localization
Satellite image classication

Top Data Types

Computer Code ProgrammingComputer Code Programming
DocumentDocument
TextText

Top Task Types

Audio Recording
Emotion Recognition
Prompt Response Writing SFT
RLHF
Translation Localization

Freelancer Overview

With extensive experience in data labeling, AI training data, and coding, I have contributed to multiple projects at Outlier AI, including Cypher RLHF, cabbage_patch, nexus_generation, and multilingual_centralized_screening, etc. My expertise spans bilingual English-Chinese translation, crafting, analyzing, and scoring prompts and responses, as well as comparing responses to refine model outputs. Additionally, my coding skills allow me to streamline workflows, optimize processes, and contribute technical solutions to complex tasks. This combination of technical and linguistic capabilities ensures high-quality contributions to multilingual AI systems. My detail-oriented approach and versatile skill set make me a standout professional in the field.

Entry LevelEnglishChinese Mandarin

Labeling Experience

Multilingual Centralized Screening

Internal Proprietary ToolingTextText SummarizationTranslation Localization
This project prioritized cross-language consistency and quality in multilingual content. Tasks included translating and localizing textual data, scoring multilingual responses, and evaluating cross-linguistic accuracy and cultural appropriateness. It ensured high-quality outputs while supporting diverse linguistic requirements for broader AI accessibility

This project prioritized cross-language consistency and quality in multilingual content. Tasks included translating and localizing textual data, scoring multilingual responses, and evaluating cross-linguistic accuracy and cultural appropriateness. It ensured high-quality outputs while supporting diverse linguistic requirements for broader AI accessibility

2024

Nexus Generation

Internal Proprietary ToolingTextText GenerationText Summarization
This initiative involved crafting prompts designed to produce coherent and contextually rich generative responses. Tasks focused on maintaining a balance between creativity and utility, ensuring outputs aligned with predetermined guidelines. The project aimed to refine generation capabilities, emphasizing high-quality textual output for dynamic applications

This initiative involved crafting prompts designed to produce coherent and contextually rich generative responses. Tasks focused on maintaining a balance between creativity and utility, ensuring outputs aligned with predetermined guidelines. The project aimed to refine generation capabilities, emphasizing high-quality textual output for dynamic applications

2024

Cypher RLHF

Internal Proprietary ToolingTextRLHFFine Tuning
This project focused on enhancing LLMs by leveraging RLHF to align model outputs with human preferences. Tasks included creating and refining prompts, analyzing response quality, and scoring outputs based on alignment criteria. Human feedback was used to train a reward model, which helped fine-tune the LLMs to improve relevance and safety in responses. Emphasis was placed on evaluating multiple responses to ensure ethical and contextually accurate outputs for diverse applications

This project focused on enhancing LLMs by leveraging RLHF to align model outputs with human preferences. Tasks included creating and refining prompts, analyzing response quality, and scoring outputs based on alignment criteria. Human feedback was used to train a reward model, which helped fine-tune the LLMs to improve relevance and safety in responses. Emphasis was placed on evaluating multiple responses to ensure ethical and contextually accurate outputs for diverse applications

2024

cabbage_patch

Internal Proprietary ToolingTextRLHFEvaluation Rating
The cabbage_patch project is part of Outlier AI's efforts to train and refine generative AI models, enabling them to generate more accurate, reliable, and user-aligned responses. This involves evaluating and ranking AI-generated text based on its accuracy, coherence, and relevance, as well as crafting prompts and responses that challenge the AI model to perform effectively in complex scenarios. The project contributes directly to RLHF workflows, improving the AI’s capacity for safe and creative interactions. Tasks are performed remotely, using flexible in-house tools, with contributors providing detailed feedback and evaluation to enhance the AI model’s performance and adaptability

The cabbage_patch project is part of Outlier AI's efforts to train and refine generative AI models, enabling them to generate more accurate, reliable, and user-aligned responses. This involves evaluating and ranking AI-generated text based on its accuracy, coherence, and relevance, as well as crafting prompts and responses that challenge the AI model to perform effectively in complex scenarios. The project contributes directly to RLHF workflows, improving the AI’s capacity for safe and creative interactions. Tasks are performed remotely, using flexible in-house tools, with contributors providing detailed feedback and evaluation to enhance the AI model’s performance and adaptability

2024

Education

U

University of Victoria

Bachelor of Art, Economics

Bachelor of Art
2014 - 2018

Work History

S

Self-Employed

Freelance Translator & Interpreter

Remote
2019 - Present
1

100Devs/Freelance

Front-End Developer

Victoria
2021 - 2023