Freelancer profile translated to English.

Description

Engineer specialized in Data Science and Artificial Intelligence, versatile Data consultant mastering the principles and technologies of Data Engineering, Data Science and ML Engineering.

Rigorous and autonomous, with a great capacity for adaptation combined with a permanent desire to acquire new knowledge.

Industry field of expertise

Languages

French
Native or bilingual
English
Fluent

Workplace preferences

Can work on-site

Paris (up to 50km)

EssilorLuxotica
Tech lead
September 2023 - September 2024 (1 year)
Implementation of a machine failure detection solution
Context: Production of defective glass in labs due to degradation of
machine tools. Each machine produces files that we will analyze.
Understanding and evaluation of business needs
- real-time data collection
- data lake construction
- provision of predictions within 5 minutes of receiving a file.
Implementation of the architecture for real-time file processing
- Reception of CSV files in a Blob Storage
- Reception of unstructured text files in a Blob Storage
- File processing with AutoLoader Databricks + workflow
- Data provision in delta tables in Databricks
Real-time prediction solution:
-Service triggered upon file reception from a machine
-The file contains an intensity mapping that we will treat as an image.
-Benchmark of a Deep Learning model for image processing (10 layers)
-Deployment of the Deep Learning model on Unity Catalog
-Sending predictions to Azure Service Bus
EssilorLuxotica
Tech lead at
January 2023 - September 2023 (8 months)
Implementation of an ML tool to assist with absence day allocation
Context: The customer service is more solicited during certain periods. Managers do not have enough visibility to make decisions regarding team management. IRMA is a project that provides predictions on the number of incoming and outgoing calls for the coming weeks.
Understanding and evaluation of business needs

• Extraction of files from an FTP server in CSV format
• Data extraction in JSON format (API request)
• Archiving of files in delta tables
• Provision of predictions once a week (on Sunday)
• Prediction duration: 2 months
• Delivery in a CSV file
Extraction and Archiving

• Schedule Azure Data Factory workflow every morning at 2 AM
• Python algorithm for extraction from an FTP server - Python algorithm for API requests, then reading JSON - Archiving data in delta tables
Prediction

• Multi-hop architecture with feature preparation in Silver delta tables in Databricks
• Implementation of an ETL workflow in Databricks scheduled only on Sundays for prediction delivery
• Evaluation of metrics (MSE) for potential retraining if the model drifts (model lifecycle via MLflow).
• Prediction in Gold tables
• Delivery of predictions to Azure Blob Storage
Qaltom
Data Engineer / Python Developer
ENERGY AND UTILITIES
October 2021 - January 2023 (1 year and 3 months)
Paris, France
Development of a charging station for electric cars
Real-time information collection using a Django API and loading data into S3
- JSON data on charging time
- JSON data on charging duration
- JSON data on system warnings and errors
Kubernetes for Backend microservices - Application divided into several microservices to more easily scale services that will be more in demand.
- Development of the charging scheduling service
- Development of the energy/cost charging report service
Extraction, Transformation, and Loading via PySpark into RDS and DynamoDB
- Extraction of JSON data from blob storage
- Data transformation to link it to real-time KW cost
- Load into DynamoDB for customer warnings (if they exceed estimated costs for their charge)
- Load customer information such as monthly charging duration into RDS