1. Established partnership with Auchan to analyze over 100GB of retail time-series data (sales data from 45k SKUs over 2 years in 50 stores)2. Predicted daily demand for 45k SKUs using ARIMA on 350 million rows of data3. Deployed development environment on GCP (Google Cloud Platform) using Python and JupyterI reached out to Data Science teams of companies to offer assistance on Data Science projects in exchange for real world industry data. Auchan, a French retailer, agreed.In a team of 6 UC Berkeley students, we built a prediction model for Auchan’s Italian stores: our goal is to predict daily demand for 45k products using data from 50 stores over the past 2 years (which amounts to 350 million rows of data and 100 GB of CSV files).Developed a model to estimate whether the sales reflect true demand, or if some products were out of stock, resulting in less sales. We do not have data on inventory, however we believe that such model would improve our demand prediction model by removing non-relevant data.Our models are built with Python (Pandas and Sklearn, Jupyter Notebook) on Google Cloud (to handle the large dataset)

1. Reviewed literature on existing work and documented EDA on Neo4j graph database (150GB)2. Contributed to the generation of ETL pipeline & feature generation3. Scaled semi-supervised model to 300MM transactions using PySpark on AWS EMR/S3 datalakeTechnology used: Neo4j / Cyper Graph databases (for handling the bitcoin database), Python (Pandas, sklearn on Jupyter Notebooks - models built and hosted on AWS EC2 instance), Apache Spark (PySpark on Zeppelin notebooks, AWS EMR on S3 datalake)Flagging suspicious activity on the Bitcoin blockchain so financial institutions can handle cryptocurrencies while ensuring their compliance. We focused on flagging transactions that went through services called mixers (which takes money from various customer and redistribute them to increase anonymity). Those services are typically used for money laundering.

1. Defined and drove product roadmap for skincare brands Decléor and Carita.2. Collaborated with engineers, product marketers and designers to design and manufacture skin-care products.3. Reduced unit cost of samples by 45% using cost-benefit analysis of procurement and supply chain options.4. RPA: Automated financial reports and product benchmarks that are reviewed by product leads and brand executives on a weekly basis, implemented on Excel in VBA.As the point of contact between the Marketing and the Industry, I managed the joined effort of the Packaging Engineers, Compatibility Lab, Production, Supply Chain, Suppliers and the Procurement Department. My goal was to make sure the products would come out a the right time, the right price and the quality set by the MarketingMy main projects included:• Managing the renovation of Decléor and Carita sampling catalogue:Reduced the unit cost of samples to be able to give out more samples in retail and e-retail & to raise brand awareness.Worked closely with the Suppliers, packaging engineers and buyers to find solutions.Results: - Handed an on-track project to my manager,- Reduced the unit cost of the samples by 31-78% (on average, 45%)• Automation of the team's benchmarking tool:Decléor and Carita are not integrated in the group's systems. Until now, the team had to manually search for products in a database to get a price decomposition of existing productsAs a temporary solution before the brands would get integrated, I decided to create an Excel tool that would search automatically give the price decomposition when given a product number (use of VLOOKUP and a macro to consolidate two databases and get the results).I primarily worked on Microsoft Excel (including VBA). I also learned how to use L'Oréal's version of TeamCenter, Minds (Product Lifecycle Management tool).Skills developed: Project Management, Stakeholder management, Effective Communication, Negociation, Data Analysis, Excel, VBA.

Managed various continuous improvement projects in Essilor's lens production plant of Antony.1. Identified, prioritized and tested process modification to reduce waste. Analyzed and presented results from the tests to the plant's managing team, and implemented promising ones. Communicated to corporate teams about the results (Europe's and Asia's tech platforms in Créteil, France, and Bangkok, Thailand)Results: 42% fewer lens rejected for cosmetic defects and 45% reduction in polishing slurry consumption (~60 000$ saved per year over the $200,000 annual budget).2. Reorganized quality check workstation for lenses before coating. Created and led a focus group composed of shop floor workers and engineers to design an efficient solution that workers would adopt.Results: both the managing team and the workers were satisfied with the solution. The focus group became champions for implementing the solution and manage the change.3. Created and edited SQL scripts to generate reports on daily operations and quality control.

I was elected President to lead and manage the team of 12 students. As team leader, I kept track of the club goals and helped members to carry on the club duties, I organized various gaming events, tech talks and training sessions on web-related technologies, and handled the club's communication strategy within the campus.The KI is in charge of the network for over 400 students living on campus, organizes gaming events every month, develops an internal website for student events (uPont) and holds conferences on subjects such as web security, Git and LaTeX.

Assisted an IT project manager in identifying the Sanitation Department's needs for management systems. To do so, I worked with sanitation workers in their field and office tasks to get a precise idea of their needs and their ease with software. After understanding and capturing their needs, I conceived a temporary Excel-based solution to improve their operations management. The City Hall is now calling for tenders for development of the final product.