Siham Chaara

Key Skills

Software

Remotasks

Scale AI

Top Subject Matter

French

Image Generation

Codes Output

Top Data Types

Audio

Image

Text

Top Task Types

Classification

Evaluation Rating

Prompt Response Writing SFT

Freelancer Overview

During my time as an AI Trainer at Outlier, I was responsible for enhancing the performance and accuracy of machine learning models. This involved not only training the models, but also critically analyzing their interactions to identify areas for improvement. I provided key recommendations for algorithmic adjustments, ensuring a continuous refinement of outputs to achieve higher precision. What differentiates me is my practical experience in AI training, underpinned by strong analytical capabilities. My work required a deep understanding of algorithmic processes and data-driven decision-making. Additionally, my organizational skills and proficiency in multiple languages, including French, English, and Spanish, enable me to navigate diverse, cross-functional projects with ease.

Labeling Experience

Workspace Evaluation

Scale AITextQuestion AnsweringEvaluation Rating

Evaluate the quality of two responses generated by AI models. Assess each response based on various quality criteria. Subsequently, determine which of the two responses is superior and provide your rationale. A unique aspect of this project is that we possess not only the prompt but also supplementary context, including the Active File/Emails referenced by the user while composing the prompt.

Evaluate the quality of two responses generated by AI models. Assess each response based on various quality criteria. Subsequently, determine which of the two responses is superior and provide your rationale. A unique aspect of this project is that we possess not only the prompt but also supplementary context, including the Active File/Emails referenced by the user while composing the prompt.

2024

Cocoa

Scale AITextQuestion AnsweringEvaluation Rating

We have the response from an AI chatbot to a prompt given by a user. This response is split into sentences (spans), and the goal is to evaluate the factual accuracy of each of those spans by looking for web resources that support or contradict what is stated by the model.

We have the response from an AI chatbot to a prompt given by a user. This response is split into sentences (spans), and the goal is to evaluate the factual accuracy of each of those spans by looking for web resources that support or contradict what is stated by the model.

2024

Onion Dance

Scale AIImageClassificationEvaluation Rating

Determine which image is the best edited given the instruction given by the user.

Determine which image is the best edited given the instruction given by the user.

2024 - 2024

Bulba Languages

Scale AITextEvaluation RatingPrompt Response Writing SFT

Evaluating AI-generated responses, particularly focusing on writing quality, instruction following, verbosity, truthfulness, and safety. The goal is to assess how well an AI response meets the desired criteria and adheres to best practices in communication.

Evaluating AI-generated responses, particularly focusing on writing quality, instruction following, verbosity, truthfulness, and safety. The goal is to assess how well an AI response meets the desired criteria and adheres to best practices in communication.

2024 - 2024

Education

Work History