SENONI, Paris / New York
05 / 2015 – now, Senior Data scientist
SENONI: data-driven consulting.
I created SENONI to apply and advise on state-of-the-art advanced analytic and quantitative tools, modeling techniques, and data mining procedures to derive business insights, solve complex business problems and improve clients decisions and the impact of key operational processes.
• 2016/02 - now: ARCELORMITTAL
- Building a recommendation system to help ArcelorMittal decide which products to sell on a particular week and which sales to postpone to the following weeks, in order to maximize revenues.
- Predicting prices per item and Customers' behaviour next week.
- Increased Sales Turnover by 8%.
- Python - MongoDB - Gradient Boosting Trees / Theano / Lasagne - Advanced feature engineering, Customized Objective Function implementing business rules - Pymc3 - A/B testing.
• 2016/07 - now: CARREFOUR
- Building a recommendation system to set daily the optimal price for online products, reducing "per item" cannibalization and maximizing margins next day.
- Python - Multi-armed bandits - Spark 2.0 - Graphframes - Bayesian Deep Learning / PyMC3
• 2015/05-09: EMOZIA: enabling products and services to understand and respond to human emotion.
Designing, developping and scaling an intelligent system to incorporate emotive recognition in apps, brands and services (Trend Analysis, Behavior Modeling, ...).
- Benchmarking Hadoop/Hive, Amazon EC2/DynamoDB/Redshift/Kinesis, Cassandra, Spark to set up the appropriate platform
- Writing more than 3000 lines of code in Python (sklearn, pandas, gensim, various APIs) to generate features from collected data, train supervised and unsupervised models to extract or predict labels such as fatigue, stress, sleep patterns, physical activity (Latent Dirichlet Allocation, Neural Networks, Hidden Markov Model, Random Forest, etc) and building an API to provide the enriched data back to customers.
- Specifying and validating Android and iOS SDKs to collect sensors data from mobile phones.
• 2016/01 - now: ********
- Image Classification with convolutional neural networks (Google TensorFlow)
- Elaborating an online tutorial to build a Data Lake and perform Machine Learning with Hadoop, Spark and Python / Scala. Packaging a standalone Development / Hacking platform including packages such as Spark, Hadoop, Hive, Impala, Maven, Scala, intelliJ IDEA, Python, Jupyter, R, Zeppelin and Git. Enriching the platform with machine learning tutorials written in Python and Scala and enabling the platform to run in standalone mode the Datacruncher developed in Spark/Scala to feed the enterprise Data Lake.
Metis, New York / USA
11 / 2014 - 04 / 2015, Data scientist
Metis is an immersive 12 weeks data science bootcamp in New York City, with curriculum design and full-time instruction by Datascope Analytics. Through five data projects, Metis teaches and reinforces concepts such as statistical inference, data acquisition, supervised and unsupervised learning, D3.js, SQL and NoSQL databases, and Hadoop MapReduce:
- Final Project: Driver Telematics Analysis - identifed trips which are not from the main driver of a given car using the data of GPS loggers and categorized drivers by "Driving Style" to build a customized insurance offer.
transition management (Renault, Yanmar, DCNS, PSA, etc.), France & Russia
05 / 2009 - 10 / 2014, Director of Complex or International Projects
- Purchasing Director: set up from A to Z a Procurement department in the French subsidiary of Yanmar - an industrial Japanese group building excavators for the European market.
- Automotive Project Director: lead the industrialization of a Korean vehicle in the Russian plant of Renault - a French car manufacturer (based in France and Russia).
- Supply Chain Management: enhanced the logistics flows of a major maintenance contract in the Marine Defense Industry (DCNS).
- Automotive Suppliers Developpement in Russia (Avtovaz supplier network to reach Renault's requirements).
- Small Business Acquisition in France - from the Letter of Intent to the Preclosure Meeting (audited target: 100-employee engineering services company).
RENAULT, Paris / France
05 / 2003 - 11 / 2009, Project Purchasing Manager
Managed 3 sections within the Renault-Nissan-Purchasing-Organization:
- Upstream Purchasing of Automotive Parts: in charge of investigating new suppliers for automotive parts, in cost effective markets such as Russia, Tunisia, Bulgaria, Slovakia and Saudi Arabia. Selected and appointed suppliers in areas as diverse as casting, forging, plastic injection, wiring, etc.
- Information Services Purchasing: supervised the IS Purchasing progress plan of the outsourcing contract with ATOS ORIGIN (software development, budget € 80 M). Negotiated the renewal of the SAP contract (20 M €) and restructured the panel of service providers in the HRIS area.
- Headed the B2B Information Systems: construction and deployment of the Renault Supplier Portal.
PHOENICIA VALLEY, Paris / France
05 / 2000 - 04 / 2003, Start-up Founder
- Established a Web Agency specialized in B2B marketplaces (php/mysql development).
- Managed all aspects of company's operations as well as handled financing/fund raising; bagged the initial 5 contracts.
PROCTER & GAMBLE, Cincinnati / USA
07 / 1998 - 04 / 2000, Internet Project Manager
Internet Consultant for the Fabric & Home Care Brand Managers, animating Interactive Marketing and Direct to Consumer initiatives in the North American market.
SUN MICROSYSTEMS, Middle East & Africa
09 / 1994 - 08 / 1997, Systems / Sales Engineer
- Installation and administration of Unix workstations, IP networks, Oracle databases.
- Tendering for the first Internet Service Providers; marketing and selling Intranet and Extranet turnkey infrastructures to Government and Education (expertise, architecture, negotiation).