For employers

Hire this AI Trainer

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

Invite to Job
Djati Mélan

Djati Mélan

Expert in AI Evaluation in fench and english.

France flagArgenteuil, France
$30.00/hrIntermediateAppenClickworkerOther

Key Skills

Software

AppenAppen
ClickworkerClickworker
Other

Top Subject Matter

No subject matter listed

Top Data Types

AudioAudio
TextText

Top Task Types

Audio Recording
Data Collection
Evaluation Rating
Relationship

Freelancer Overview

My first project involving AI focused on testing the limits of a large language model (LLM). The goal was to design prompts that would cause the model to fail, analyze those failures across different dimensions, and then rewrite the responses to correct the issues. Later, I was also responsible for validating the work of others, ensuring consistency and accuracy across the team’s outputs. This required a sharp eye for detail, critical thinking, and a deep understanding of how the model processes information. I approached the task analytically and documented each weakness to support improvements in the model’s performance. This pattern-recognition skill is something I consistently apply across other areas of my work. In trading, in sports analysis (where my role involves studying and anticipating opponent behavior), and in everyday interactions, I’ve developed a strong ability to detect and leverage patterns. That same mindset helped me quickly identify flaws in the LLM and approach the task methodically. I’ve also contributed to more conversational AI projects, both written and spoken, expanding my practical experience with AI in real-world contexts.

IntermediateFrenchEnglish

Labeling Experience

Project Vertex Audio English (french accent)

OtherAudioAudio Recording
Project goal : This project aims to stress-test AI performance in realistic conversational environments, evaluating AI’s capability to handle challenging audio conditions during human interactions. Your task is to have a conversation with an AI in diverse backgrounds including some conversations in a quiet room and a loud environment.

Project goal : This project aims to stress-test AI performance in realistic conversational environments, evaluating AI’s capability to handle challenging audio conditions during human interactions. Your task is to have a conversation with an AI in diverse backgrounds including some conversations in a quiet room and a loud environment.

2025 - 2025

Project Vertex Audio French

OtherAudioAudio Recording
Project goal : This project aims to stress-test AI performance in realistic conversational environments, evaluating AI’s capability to handle challenging audio conditions during human interactions. Your task is to have a conversation with an AI in diverse backgrounds including some conversations in a quiet room and a loud environment.

Project goal : This project aims to stress-test AI performance in realistic conversational environments, evaluating AI’s capability to handle challenging audio conditions during human interactions. Your task is to have a conversation with an AI in diverse backgrounds including some conversations in a quiet room and a loud environment.

2025 - 2025

cypher_evals_c

OtherTextEvaluation Rating
Validating the work of others, ensuring consistency and accuracy across the team’s outputs.

Validating the work of others, ensuring consistency and accuracy across the team’s outputs.

2024 - 2024

cypher_rlhf

OtherTextPrompt Response Writing SFT
The goal was to design prompts that would cause the model to fail, analyze those failures across different dimensions, and then rewrite the responses to correct the issues. Each prompt had to fit within an assigned category (such as summarization, classification, rewriting, creative writing or others, as it was frequently updated) and adhere to a specific topic (e.g., biology, education, etc.)

The goal was to design prompts that would cause the model to fail, analyze those failures across different dimensions, and then rewrite the responses to correct the issues. Each prompt had to fit within an assigned category (such as summarization, classification, rewriting, creative writing or others, as it was frequently updated) and adhere to a specific topic (e.g., biology, education, etc.)

2024 - 2024

Education

U

Université de Cergy-Pontoise

Bachelor's Degree, Economics And Management

Bachelor's Degree
2024 - 2024
I

Inseec

Master's Degree, Digital Strategy And Project Management

Master's Degree
Not specified

Work History

G

Google

Coach Hotline

N/A
2021 - 2024
A

Adforall

Account Manager

N/A
2019 - 2020