Stephen S.

Key Skills

Software

No software listed

Top Subject Matter

LLM and Speech-to-Speech model evaluation and AI data annotation

Top Data Types

Text

Audio

Top Task Types

Tracking

Transcription

Freelancer Overview

OUTLIER | Spécialiste Évaluation & Entraînement de Modèles IA (LLM / Speech-to-Speech). Brings 17+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Outlier Platform. Education includes Master of Science in Computer Science, Paris 12 University (2011) and Bachelor of Science in Computer Science, Université d'Orléans (2009). AI-training focus includes data types such as Text and Audio and labeling workflows including Evaluation, Rating, and Tracking.

Labeling Experience

OUTLIER | Spécialiste Évaluation & Entraînement de Modèles IA (LLM / Speech-to-Speech)

Text

Annotation and evaluation of data to improve LLM and Speech-to-Speech model outputs through structured quality and safety assessment. You perform comparative rating of generated responses using criteria such as quality, security, relevance, and fluency, following Side-by-Side/ELO methodologies. You also evaluate Live Speech-to-Speech streams by scoring linguistic accuracy, tone, and naturalness, and write detailed FR/EN justifications to guide algorithm improvements. • Comparative LLM response evaluation (Side-by-Side / ELO Ranking) • Speech-to-Speech live evaluation with scoring of accuracy/tone/naturalness • Creation of detailed analytical rationales in FR/EN • Quality improvement of training data via rigorous, consistent output labeling

2026 - Present

AI Model Evaluation Specialist - Outlier

AudioTrackingTranscription

You performed comparative evaluation of language and speech-to-speech model outputs to support improvements in model quality and safety. You assessed generated responses against criteria such as quality, security, relevance, and fluency and provided detailed FR/EN justifications to guide engineering adjustments. You also evaluated speech-to-speech live flows by scoring linguistic accuracy, tone, and naturalness. • Conducted Side-by-Side and ELO-style comparative evaluations • Rated live speech-to-speech outputs for linguistic precision and naturalness • Wrote detailed analytical feedback and justifications in FR/EN • Contributed to consistent annotation quality for training data improvement

2026 - 2026

Education

Work History