Freelancer profile translated to English.

Description

Specializing in LLMs, machine learning, generative AI, and MLOps, I help companies design and deploy high-impact data-driven solutions. 5 years of experience transforming complex business problems into robust and scalable predictive models.

Languages

French
Native or bilingual
Arabic
Native or bilingual
English
Fluent
Italian
Native or bilingual

Workplace preferences

Can work on-site

Paris (up to 50km)

Kantar
LLM Engineer | MLOps & Generative AI
November 2024 - Today (1 year and 9 months)
Paris, France
Fine-tuned and optimized specialized LLMs (CamemBERT, Mistral, GPT-4) to automate customer feedback analysis — reduced manual processing time by 70% while increasing insight accuracy to 92%, enabling real-time monitoring of customer sentiment in the French market.
Designed and deployed RAG systems with multi-modal LLMs for intelligent document retrieval — accelerated research and analysis cycles by 5x, allowing consultants to handle 3x more client projects while maintaining 99% data accuracy for critical industry decisions.
Designed and trained ensemble predictive models (XGBoost, Transformers) to forecast consumer behavior — achieved 85% accuracy in predicting quarterly purchasing trends, directly increasing customer campaign ROI by 15% through hyper-targeted marketing strategies.
Built and automated enterprise MLOps pipelines with model optimization — reduced deployment cycles from 2 weeks to 3 days and decreased cloud infrastructure costs by 25%, enabling rapid AI model iteration across multiple client portfolios.
Optimized LLMs and implemented ethical AI practices by testing quantization techniques and detecting biases — reduced inference time by ~40% and model size by ~20%, while improving fairness across key demographic segments for more reliable AI deployments on client projects.
Python (Programming Language) Generative AI MLOps Mistral AI Amazon Web Services
CSTB
LLM Engineer | MLOps & Generative AI
ARCHITECTURE AND URBAN PLANNING
September 2022 - September 2024 (2 years)
Paris, France
Automated complex environmental data processing by developing scalable data pipelines, reducing manual data processing time by 30% and enabling real-time monitoring of air quality metrics across multiple facilities.
Applied unsupervised learning algorithms (K-means, Hierarchical Clustering) to identify hidden patterns in environmental data, resulting in a 20% improvement in detecting abnormal pollution levels and enabling proactive interventions.
Designed a cloud infrastructure on AWS (S3, EC2) for large-scale data storage and processing, reducing storage costs by 25% while improving data accessibility for cross-functional teams.
Initiated LLM applications for automated analysis of occupant survey data, reducing analysis time by 40% and improving accuracy in identifying key comfort and health issues from qualitative feedback.
Identified 3 distinct pollution profiles using advanced clustering techniques, enabling targeted ventilation strategies that reduced pollutant concentrations by 35% in high-risk areas of healthcare facilities.
Provided data-driven renovation recommendations implemented in 15+ healthcare facilities, leading to a 28% improvement in indoor air quality and better patient recovery conditions.
Facilitated evidence-based decision-making by translating technical insights into executive reports, resulting in a 50% faster approval process for environmental improvement projects and budget allocations.
Python (Programming Language) AWS Cloud AI Agent Generative AI MLOps
CSTB
LLM Engineer – Applied AI & Environmental Data Analytics
ARCHITECTURE AND URBAN PLANNING
September 2021 - August 2022 (11 months)
Paris, France
Conducted air quality analyses in over 500 French schools for CSTB programs, improving the detection of non-compliant buildings by 35%.
Developed environmental analysis models (predictive models for CO₂, anomaly detection, classification models, PCA, clustering), reducing false positives in IAQ diagnostics by 40%.
Automated the interpretation of technical reports using GPT-based LLMs, reducing manual analysis time by 70%.
Built an end-to-end data pipeline for sensor data ingestion, cleaning, and aggregation (Python, SQL, Airflow, Docker), improving data consistency by 30%.
Designed an IAQ scoring system combining indicators for CO₂, VOCs, PM2.5, humidity, and ventilation, adopted by over 50 local authorities to prioritize interventions.
Developed Power BI dashboards for real-time IAQ level monitoring and faster response to threshold exceedances, widely adopted by field teams.
Python (programming language) Machine Learning MLOps artificial intelligence AI Chatbot