For employers

Hire this AI Trainer

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

LoginCreate account
Invite to Job
Alexandra Keller

Alexandra Keller

Content writer and editor and business savvy freelancer

Switzerland flagGeneva, Switzerland
$40.00/hrEntry LevelInternal Proprietary ToolingOther

Key Skills

Software

Internal/Proprietary Tooling
Other

Top Subject Matter

No subject matter listed

Top Data Types

TextText

Top Task Types

Prompt Response Writing SFT
RLHF
Translation Localization

Freelancer Overview

I have hands-on experience in data labeling and AI training through my work on the Crypher RLHF project with Outlier, where I contributed as a French (Swiss) language expert. This project focused on Reinforcement Learning with Human Feedback (RLHF), aimed at enhancing AI's ability to understand and respond accurately to diverse prompts across a range of categories and subjects. My role involves generating high-quality prompts tailored to specific contexts and evaluating AI responses based on key criteria such as Instruction Following, Localization, Writing Quality, Verbosity, and Truthfulness. This meticulous evaluation ensured that the AI delivered precise, culturally nuanced, and contextually appropriate outputs. What sets me apart is my fluency in French with a deep understanding of Swiss linguistic and cultural nuances, combined with a keen eye for detail and commitment to accuracy. I excel at assessing subtle linguistic and contextual variations, enabling me to provide meaningful feedback that improves AI performance. My ability to balance creative prompt generation with rigorous evaluation standards has been instrumental in training AI systems that are more reliable, responsive, and aligned with user expectations.

Entry LevelFrenchEnglish

Labeling Experience

Cypher rlhf

OtherTextRLHF
Participants are guided through the process of creating conversations with a chatbot. The task begins with drafting a prompt, which serves as a starting point for generating two responses from the AI model. The overarching goal is to craft prompts that lead to one of the responses exhibiting a Model Failure (an error or inadequacy in the AI's output). If neither response demonstrates such a failure, the prompt needs to be adjusted and refined to provoke this situation. Once the responses are generated, each one is evaluated based on seven specific attributes, such as accuracy, humor, conciseness, and more. Participants then choose the better response of the two and provide a justification for their choice. The final step involves rewriting the selected response to make it flawless—addressing issues like truthfulness, tone, brevity, or any other identified weaknesses.

Participants are guided through the process of creating conversations with a chatbot. The task begins with drafting a prompt, which serves as a starting point for generating two responses from the AI model. The overarching goal is to craft prompts that lead to one of the responses exhibiting a Model Failure (an error or inadequacy in the AI's output). If neither response demonstrates such a failure, the prompt needs to be adjusted and refined to provoke this situation. Once the responses are generated, each one is evaluated based on seven specific attributes, such as accuracy, humor, conciseness, and more. Participants then choose the better response of the two and provide a justification for their choice. The final step involves rewriting the selected response to make it flawless—addressing issues like truthfulness, tone, brevity, or any other identified weaknesses.

2024

Education

E

EU Business School

Bachelor, Business Administration

Bachelor
2018 - 2020

Work History

A

Alexandra Keller

Content writer & editor

Geneva
2021 - Present
A

Alexandra Keller

Project Manager

Geneva
2021 - Present