Freelancer profile translated to English.

Description

With over 6 years of experience in data, I combine a dual expertise in Data Engineering and Data Science, ranging from the design of robust pipelines to predictive modeling. I work across the entire value chain, from data ingestion to exploitation, ensuring the industrialization of solutions in complex environments. My product-oriented approach and my ability to coordinate technical and business teams enable me to successfully lead high-impact data projects.

Languages

French
Native or bilingual
English
Native or bilingual

Workplace preferences

Can work on-site

Paris (up to 50km)

SNCF,
AI Engineer
March 2021 - August 2022 (1 year and 5 months)
Paris, France
Participation in strategic projects focused on user data valorization and large-scale analytical flow industrialization.

Development of an intelligent OCR:Design of an automated data extraction pipeline from large PDF files. Selection of Google Vision API after benchmarking, integration of OpenCV pre/post-processing, and automatic generation of structured Excel files with quality controls. The tool was industrialized and reused in several cases.

Behavioral Analysis & ML:Construction of an ETL pipeline on Redshift to aggregate, clean, and enrich application usage data. Calculation of behavioral KPIs (frequency, errors, satisfaction), modeling with K-Means, evaluation via Silhouette/Davies-Bouldin. Results were integrated into Power BI for interactive weekly analysis.

STT Project with Notion API:Development of a voice agent based on Whisper and LangChain/LangGraph. Modular LLM architecture with RAG system for contextual enrichment and automatic synchronization in Notion. Tool used to capture user feedback.

AWS Data Lake:Implementation of a multi-layer Data Lake (Landing, Bronze, Silver) with Glue, Step Functions, Redshift. Secure connections to third-party APIs, quality management with CloudWatch logs and SNS alerts, unit tests on critical steps to ensure robustness and scalability.

OPTI RUN & USERS ANALYTICS:Creation of star schema Gold DataMarts, integration of business management rules. Industrialization via Glue/Step Functions, supervision via CloudWatch/SNS. Complete automation of processing, quality monitoring, and support for Power BI dashboards.

Environment: AWS (Glue, S3, Redshift, Lambda, Step Functions, CloudWatch, SNS), Python, SQL, Power BI, Scikit-learn, OpenCV, LangChain, Whisper, Notion API.
Python SQL Data science intelligence artificielle Data Engineer
Mangabey
Lead Data Engineer
MEDICAL
September 2022 - December 2023 (1 year and 3 months)
Paris, France
Design and orchestration of AI solutions for large-scale pharmaceutical data valorization:
• Management of a team of data engineers and technical coordination.
• Deployment of robust ETL pipelines for unifying and processing multi-source and multi-country data.
• Development of ML/DL algorithms for data completion and reliability.
• Integration of LLM models for automatic standardization of heterogeneous data.
• Client interface and Agile management to ensure the quality and impact of deliverables.
ADP - LISSI Lab.,
AI Researcher & Engineer
March 2021 - August 2022 (1 year and 5 months)
Paris, France
Participation in several advanced AI research and engineering projects, conducted in partnership with institutions such as Paris Airports (ADP), focusing on human activity recognition, contextual AI, image processing, and medical diagnosis assistance.

🔹Human Activity Recognition (ADP)

Implementation of an experimental protocol with REST APIs to collect multi-sensor data.

Development of an ETL pipeline for processing, synchronizing, and enriching data streams.

Creation of labeled datasets, training of Deep Learning models to classify human activities.

🔹Hybrid Contextual AI (Deep Learning + Logical Reasoning)

Modeling of semantic ontologies (activity, location, temporality) and integration of logical inference engines.

Design of an intelligent system combining Deep Learning and logic to detect complex contextual situations.

🔹Computer Vision for Autonomous Robotics

Creation of an object dataset (RGB camera), labeling, and training of CNNs for detection and classification.

Deployment of a containerized REST API under Docker for real-time robotics integration.

🔹NLP & Intelligent Voicebot (GPT-2 + Ontology)

Design of an STT (Speech-to-Text) chain with contextual understanding.

Structuring of intents in an ontological database and integration of an enriched GPT-2 model for complex dialogues.

🔹Parkinson's Disease Diagnostic Aid Tool (Medical Project with LAMA)

Extraction and processing of biomechanical signals (vGRF) from underfoot sensors.

Statistical analysis, clustering, outlier detection, visualization in Power BI.

Development of an unsupervised algorithm based on mixed polynomial regression models (PRM).
Result: 91% accuracy, tool integrated into the lab, validated by scientific publication.