Bruno Decarnoncle

Key Skills

Software

Data Annotation Tech

Top Subject Matter

No subject matter listed

Top Data Types

Computer Code Programming

Image

Text

Top Task Types

Computer Programming Coding

Evaluation Rating

Text Generation

Translation Localization

Freelancer Overview

I consistently deliver high-quality work with strong attention to detail, even on long tasks. My reliability and top performance led to a quick promotion as a data labeling reviewer. I’m thorough, fact-check carefully, assess text naturalness, evaluate code performance, and refine AI outputs to ensure the highest quality. I’ve worked on a wide range of projects across different domains. In Belgian French localization, some of the tasks I’ve handled include creative writing, open and closed Q&A, text summarization, and structured text generation, just a few examples of the linguistic work I’ve contributed to. In math, I’ve worked on problems involving geometry, calculus, algebra, and word problems, among other topics. On the coding side, I’ve focused on .NET-related projects, such as test generation, debugging, solution exploration, and code generation, bringing both accuracy and deep technical insight to each task.

Labeling Experience

AI Tutor

Data Annotation TechImageObject DetectionText Summarization

This set of ongoing short-term projects focused on developing and evaluating multimodal AI models that combine image understanding with text interaction in Belgian French. The project was divided into two key parts. The first involved sourcing high-quality, royalty-free images that matched specific, culturally relevant prompts, ensuring each image was contextually rich. The second part focused on engaging in multimodal conversations with the models, where I evaluated and corrected two models’ responses based on both visual and textual inputs. Tasks included rating model answers for relevance and naturalness, identifying missing visual context or misunderstandings, and summarizing the intended meaning for clearer output generation. Accuracy, cultural appropriateness, and linguistic clarity were key quality benchmarks across all contributions.

This set of ongoing short-term projects focused on developing and evaluating multimodal AI models that combine image understanding with text interaction in Belgian French. The project was divided into two key parts. The first involved sourcing high-quality, royalty-free images that matched specific, culturally relevant prompts, ensuring each image was contextually rich. The second part focused on engaging in multimodal conversations with the models, where I evaluated and corrected two models’ responses based on both visual and textual inputs. Tasks included rating model answers for relevance and naturalness, identifying missing visual context or misunderstandings, and summarizing the intended meaning for clearer output generation. Accuracy, cultural appropriateness, and linguistic clarity were key quality benchmarks across all contributions.

2024

AI Tutor

Data Annotation TechTextText GenerationText Summarization

In this ongoing series of short-term projects, I contributed to developing bots capable of understanding and generating content in Belgian French, with a focus on culturally relevant context and accurate localization. Tasks included summarizing long-form content into clear, concise answers, translating and localizing English text into native-level Belgian French, and generating Q&A data based on various prompts. I was also responsible for evaluating and rating AI outputs for fluency, accuracy, and naturalness. A unique part of this project involved working on .NET programming-related content. I labeled and generated code examples, corrected syntax or logic issues, and reviewed AI-written explanations to ensure technical accuracy and clarity, drawing on my experience as a senior .NET developer. Each task was completed with close attention to language nuance, logic consistency, and adherence to project guidelines.

In this ongoing series of short-term projects, I contributed to developing bots capable of understanding and generating content in Belgian French, with a focus on culturally relevant context and accurate localization. Tasks included summarizing long-form content into clear, concise answers, translating and localizing English text into native-level Belgian French, and generating Q&A data based on various prompts. I was also responsible for evaluating and rating AI outputs for fluency, accuracy, and naturalness. A unique part of this project involved working on .NET programming-related content. I labeled and generated code examples, corrected syntax or logic issues, and reviewed AI-written explanations to ensure technical accuracy and clarity, drawing on my experience as a senior .NET developer. Each task was completed with close attention to language nuance, logic consistency, and adherence to project guidelines.

2024

AI Tutor

Data Annotation TechAudioAudio Recording

In this series of audio-focused projects, the goal was to help improve the model’s ability to process spoken instructions across various topics and environments. I recorded voice instructions in native Belgian French, covering a range of prompts related to everyday Belgian context and .NET programming concepts. Each recording session involved giving clear, structured commands that a model could interpret and act upon. To enhance the robustness of the model, recordings were done in varied acoustic environments with different types of background noise, such as household sounds, street ambiance, or quiet indoor settings. This helped train the model to handle real-world audio complexity while maintaining accuracy. A diverse mix of prompt categories and noise profiles was used to ensure broad model coverage and improve performance in natural, noisy conditions.

In this series of audio-focused projects, the goal was to help improve the model’s ability to process spoken instructions across various topics and environments. I recorded voice instructions in native Belgian French, covering a range of prompts related to everyday Belgian context and .NET programming concepts. Each recording session involved giving clear, structured commands that a model could interpret and act upon. To enhance the robustness of the model, recordings were done in varied acoustic environments with different types of background noise, such as household sounds, street ambiance, or quiet indoor settings. This helped train the model to handle real-world audio complexity while maintaining accuracy. A diverse mix of prompt categories and noise profiles was used to ensure broad model coverage and improve performance in natural, noisy conditions.

2024 - 2024

Education

Work History