Youssef Benaouda

Key Skills

Software

LabelImg

Label Studio

Prodigy

Top Subject Matter

No subject matter listed

Top Data Types

Audio

Image

Text

Top Task Types

Bounding Box

Classification

Object Detection

Segmentation

Text Generation

Freelancer Overview

Freelance AI engineer with strong experience in data labeling for computer vision and NLP tasks. At Aeroprickle, I contributed to developing computer vision models after annotating drone shot video footage of cactus, using tools like LabelMe, LabelImg, and Label Studio. I optimized the labeling pipeline to improve model performance and supported the training of MASK RCNN and CNN-based detection models with custom-labeled data. In another project focused on mental health detection from Speech, I built multi-output NLP classifiers by manually labeling thousands of lines of extracted text with depression, anxiety, and suicidal indicators. I used Prodigy and custom Python scripts for annotation, and leveraged scikit-learn, TF-IDF vectorizers, and Logistic Regression to train and evaluate the models.

Labeling Experience

Emergency Call Center Voice Annotation & French Transcription for Incident Detection

Label StudioAudioDiagnosisPrompt Response Writing SFT

AI project for transcribing and annotating emergency call center audio, to support real time incident detection. Annotated over 800 call recordings by transcribing french speech, labeling emotional tone (stress, panic, calm), and marking speaker turns. Classified utterances by intent (e.g. medical emergency, fire, false alarm). Data was used to train a multilingual speech-to-text model and emotion-aware classifier to assist first responders.

AI project for transcribing and annotating emergency call center audio, to support real time incident detection. Annotated over 800 call recordings by transcribing french speech, labeling emotional tone (stress, panic, calm), and marking speaker turns. Classified utterances by intent (e.g. medical emergency, fire, false alarm). Data was used to train a multilingual speech-to-text model and emotion-aware classifier to assist first responders.

2024 - 2024

Cactus Images Cladodes Object detection

LabelimgImageBounding BoxSegmentation

AEROPRICKLE is a client that needed help with developing computer vision models for insect detection. I annotated over a thousand lines datasets using LabelImg, and Label Studio. I worked on different use cases, had to use bounding boxes, polygon masks as well as Landmark annotation

AEROPRICKLE is a client that needed help with developing computer vision models for insect detection. I annotated over a thousand lines datasets using LabelImg, and Label Studio. I worked on different use cases, had to use bounding boxes, polygon masks as well as Landmark annotation

2023 - 2024

Mental Health Speech Annotation for Multi-Label Classification

ProdigyTextEmotion Recognition

Annotated several thousand lines of extracted text to detect signs of depression, anxiety, and suicidal ideation from speech. Tasks included defining labeling schema, cleaning extracted text, and applying multi-label tags. Used Prodigy and Python tools to support model training with TF-IDF and Logistic Regression.

Annotated several thousand lines of extracted text to detect signs of depression, anxiety, and suicidal ideation from speech. Tasks included defining labeling schema, cleaning extracted text, and applying multi-label tags. Used Prodigy and Python tools to support model training with TF-IDF and Logistic Regression.

2023 - 2023

Education

Work History