Data consultant at Keyrus France

• AI Data Quality

  Context
  In many companies, customer databases sometimes contain anomalies, which makes the marketing part complicated and some campaigns costly. Indeed, sending messages to the wrong numbers or to the wrong email prevents the goal of informing the customer from being fulfilled. There is therefore a need to detect these anomalies and to propose corrections if correctable.
  Missions
  - First, detect anomalies in the data and return stats.
  - Second, propose an optimal way to correct them.
  
  

• Creation of a mapping between product repositories

  Context
  Our client, a major B2B player, wants to create a single repository for its products. Today, the client has many repositories in the classification of its products: 1/ In the PIM (PIM Class); 2/ In the MDM (B-Class); 3/ In the ERPs of the different countries (Local Classification). Today, this multitude of classifications makes it difficult to have a single vision of its products. The client has already started work in 2024, with the help of Keyrus, to simplify and improve the classification at the PIM level. The objective of its work is to continue its work, as part of the migration of the MDM to Informatica P360 to create a single mapping on all classifications. Missions
  - Framing and understanding the need
  - Development of data processing flows to analyze data quality and answer questions through EDA
  - Code review with another Data Scientist in order to modularize and industrialize - the code in reuse for analysis with the various local ERPs
  - Production of summary documents of iterations on EDA Participation in the preparation of the restitution session for the project's IT sponsors
  

• Accounting lettering

  Context
  Accounting lettering is an activity that allows you to reconcile (match) different invoices with their respective payments. This activity is carried out in accounting and was done manually; recently, software has also been doing it. In order to create a more efficient machine with a machine learning approach, automation of lettering is therefore proposed
  Missions
  The missions are diverse and of varying importance:
  - Understanding the algorithmic approach to be applied: understanding the backpack algorithm and adaptation to the accounting context
  - Development of the greedy approach to the method
  - Optimization of the method with a Bayesian approach
  

Data intern at Keyrus France

• Synthetic data Generation

  Context
  Our client, a major player in the mutual insurance sector, is currently recovering data from the various mutual insurance companies in France. He would like to be able to provide mutual insurance companies with datasets representative of the French population today in order to enable advanced analytical exploitation: scenario simulation, forecasting, etc. The objective is through a POC to demonstrate to them the possibilities offered by synthetic data
  Missions
  - Understanding of the need
  - Creation of an algorithm for automatic generation of synthetic data
  - Benchmark of existing solutions
  - Presentation and demonstration of the tool during customer feedback
  
  

• Hackathon dataiku

  Context
  As part of a Hackathon organized by Dataiku, the objective of this hackathon is to develop a solution using the latest Dataiku features on Generative AI and LLMs on the theme "Fan Enablement Challenge". The objectives are:
  - Use LLM Mesh functionalities
  - Develop an understanding of the architecture of LLM Mesh
  - Communicate results through social networks
  Assignment
  - Identify a dataset on which to carry out the hackathon
  - Use the Dataiku platform to analyze, explore, categorize, use LLM models, etc.
  - Work within a team of 4 people
  
  

• Tierce Maintenance Applicative

  Context
  For the Botanic customer, simple and scalable maintenance is implemented. As part of my end-of-study internship, I have to work on this client's TMA.
  Missions
  The missions are diverse and of varying importance:
  - Daily checks on the TMA (proper execution of jobs, data feedback, sending reports)
  - Flow evolution to add controls.
  

IT Manager at Lafya Proj'etude Home

Freelance IT department manager. Intervenes occasionally on the creation and maintenance of assets linked to the Lafya company

Data Science Intern at International Agency for Research on Cancer

𝗖𝗼𝗻𝘁𝗲𝘅𝘁
As part of the validation of the master 1, I did an internship in the field of data at the International Center for Research on Cancer.
𝗦𝘂𝗯𝗷𝗲𝗰𝘁
Optimal transport for automatic alignment of untargeted metabolomics data.
𝗔𝘀𝘀𝗶𝗴𝗻𝗺𝗲𝗻𝘁
• Set up a detailed restitution notebook and use of the alignment model set up.
• In R, understand the developed preprocessing code and update it to adapt it to new data.
• Verify that the developed model generalizes well to new metabolite data.

development and data internship at Kangam-Labs

𝗖𝗼𝗻𝘁𝗲𝘅𝘁
To deepen my skills in the field of development, I completed a free internship in license 2.
𝗠𝗶𝘀𝘀𝗶𝗼𝗻𝘀
• Improved skills in creating backends with SpringBoot
• Deepen full-stack development with the Symfony framework (php+twig)
• Learn a little AI with python for CV exploitation.