Laurent

For project Hyp4Uses:
- Hyperspectral images treatment
- Development of a python package for analysis automation
- Creation of a reference spectrum library
- Development of a materials classification model based on the library

For Exelsius:
- Varnish detection on PCB boards
- Images stitching
- Image analysis and thresholding to detect the application of varnish by fluorescence under UV LEDs

For CMA-CGM:
- Modelization of cancelation by client for bookings.
- Use of Graph Neural Networks on subjects such as dangerous cargo detection, ports congestion forecast.
- Modelization of port congestion in order to better predict vessels' time of arrival at berth

For Michelin Green Mobility / Watèa:
Modelization of Electric Vehicles' energy consumption based on forecoming journey.
- Extraction and analysis of vehicle CAN data
- Call to API for journey planning
- Data contextualization and enrichment (slope, urban environment)
- Modeling and forecasting
- Develop a GUI application to use predictions on real-world road journeys

Mission for UBS, Monaco:
- Train 5 employees on basics of Machine Learning (theory & application in Python)

Mission at DDI-Michelin:
- Recommendations on measures, methods and data integration for the capture and analysis of drivers' behavior

Mission at CMA-CGM:
- Container detection on images (Deep Learning algorithm based on ChainerCV to learn to detect objets « containers » in still images and videos)
- Demurage & Detention penalties forecast:
- Identify most relevant predictive variables, define a predictive model in R, push project in production by translating algorithm in PySpark on Hadoop in order to automate and speed up process, communicate results through a Qlik Sense application.

Mission for Data ScienceTech Institute, higher education data-specialized institution:
- Admission Technical interviews
- Evaluate maths & IT skills for candidates to Data Science and Data Engineering training
programs

Mission for MANE & Fils, flavour & fragrance manufacturer:
- Implementing an HR KPI system for the EMEA region
- Gather ideas from professionals to understand the need
- Detail data architecture, automate imports
- Create dashboards for different profiles (managers, HR, board members)

Mission at Massa Pneu:
- Classify customers according to their purchasing behavior, forecast acceptable price according to sale’s context and customer type

Discover new data sources and integrate them in the software through Python and MongoDb.
Look for improvements in clustering methods and in prediction algorithms.