Mélanie Bontoux

Key Skills

Software

Scale AI

Top Subject Matter

general culture

Translation/Linguistic in FR/ES/EN

Arts/cinema/Theatre/Musics/Painting

Top Data Types

Image

Audio

Text

Top Task Types

Object Detection

Fine Tuning

Evaluation Rating

Audio Recording

Classification

Question Answering

Text Summarization

Prompt Response Writing SFT

Transcription

Translation Localization

Freelancer Overview

I am an experienced AI data trainer and quality analyst with a strong background in prompt engineering, model training, data labeling, and response evaluation. My work includes auditing and enhancing contributor tasks, providing detailed feedback, and ensuring high-quality deliverables for clients in the AI and machine learning sector. I have achieved top performance ratings in quality analysis roles, and my multilingual proficiency in French, Spanish, and English allows me to work effectively on diverse, global projects. My expertise spans various domains, including NLP, translation, subtitling, and localization, and I am skilled in tools such as Hubstaff, Slack, Trados, and Aegisub. I am passionate about delivering precise, reliable training data and thrive in dynamic, remote environments where adaptability and attention to detail are essential.

Labeling Experience

Writing Analysis

Scale AITextClassificationQuestion Answering

I have to analyze different prompts/responses that are closed QA/open QA/summarization/chatbot/creative writing/classification/rewrite/extraction/etc. tasks. Depending on the prompt category, a reference text is provided, and I have to check whether the answers given by the models are correct. I have to analyze them to see if the facts are true, if the prompt has been followed, if it is well written, if there are any safety issues, etc.

I have to analyze different prompts/responses that are closed QA/open QA/summarization/chatbot/creative writing/classification/rewrite/extraction/etc. tasks. Depending on the prompt category, a reference text is provided, and I have to check whether the answers given by the models are correct. I have to analyze them to see if the facts are true, if the prompt has been followed, if it is well written, if there are any safety issues, etc.

2024

Translation

Scale AITextTranslation Localization

I had to check whether the translations made by models were faithful to the original segments and correct them if they were not.

I had to check whether the translations made by models were faithful to the original segments and correct them if they were not.

2025 - 2025

Transcription analysis

Scale AITextTranscription

I had to read meeting transcripts with several people and analyze the responses of the models who had to summarize according to certain criteria. I had to check whether all the elements mentioned in the responses were based on the actual transcript, whether any were missing, whether the facts were true, whether the meaning was preserved, etc.

I had to read meeting transcripts with several people and analyze the responses of the models who had to summarize according to certain criteria. I had to check whether all the elements mentioned in the responses were based on the actual transcript, whether any were missing, whether the facts were true, whether the meaning was preserved, etc.

2025 - 2025

Voice Acting

Scale AIAudioAudio Recording

I had to hold conversations with several people on different everyday topics, hold debates, etc., while following the instructions we were given: to speak every 6 seconds, to interrupt each other, to speak in a normal manner, to use a certain tone of voice, etc.

I had to hold conversations with several people on different everyday topics, hold debates, etc., while following the instructions we were given: to speak every 6 seconds, to interrupt each other, to speak in a normal manner, to use a certain tone of voice, etc.

2025 - 2025

Image analysis

Scale AIImageObject DetectionFine Tuning

I had to check whether the image quality was acceptable, whether everything made sense, responded well to the prompt, what needed improvement, etc.

I had to check whether the image quality was acceptable, whether everything made sense, responded well to the prompt, what needed improvement, etc.

2024 - 2024

Education

Work History