- Data Analyst
2016 - 2017
Mission: Belonging to a transversal team (eCrm), my role is to implement, integrally or partially, data science process, using many different data sources, for Marketing department:
• Implementing machine learning algorithms from scratch (collaborative filtering, recommendation system) targeted for geographic department, user segmentation, data featuring, rule based classification;
• Implementing and maintain .Net batches (one shot and scheduled)
o or marketing business requirement (mail campaigns), etc.;
• Setting up data pipelines (with SSIS) and workflows;
• Integrating data sources from Couchbase, data bases, flat files, etc. provided by the DMP (kafka, zookeeper) databases and consolidated in the Data Lake (sql server moved to hdfs in Hadoop’ cluster).
• Visual studio add-ins to mutualize internal development patterns and practices.
• Benchmarking data aggregation with Flume vs periodic data extraction.
Technical environment: C#, SSIS, sql server, Kafka, Zookeeper, couchbase, windows server, linux, .Net, Git, VS2015, PowerShell, Python, Hadoop (Cloudera), Python Tool for VS, AzureML, Azure Machine Learning Studio, Linux (CentOs).
GDF SUEZ Trading
- .Net Developper
2015 - 2016
.Net developer at GDF SUEZ Trading, La Défense, France.
Mission: Develop (within BEE entity) a new app Omega for gas asset management (Sale/Purchase, distribution, optimization, production, etc.). I’m involved as scrum team member for
• Maintain and deliver new functionalities.
• Unit and acceptance automatic tests.
• Integrating in an internal formula computation engine new math functions.
• Dev and Integration environment app delivery.
Technical environment: C#, WPF (MVVM), sql server 2008 R2, T-SQL, WCF, Moq, TFS 2010, VS 2010/2015, PowerShell.
- .Net & Hadoop Developper
2014 - maintenant
RIA & RDA .Net developer : Silverlight, WPF (MVVM); ASP.Net ( MVC 5), RIA & Data Services.
Hadoop developer : windows / linux environments : Azure, HDInsight, powerShell, Hive, Pig, Hadoop, Map Reduce (C#, java) eclipse, Visual studio, MRUnit, Combiner, Partitionner.
Mutuelle des Architectes Français
- .Net developer
2013 - 2015
In the larger context of decommissioning the old information system, I was enrolled as a .Net developer in two .Net projects (intra and extra net apps) for selling insurance contracts.
My tasks were
1/ to maintain and deliver new functionalities of the above apps as team member, considering technical constraints such as web browser compatibility (IE 7, 8, 9, 10, 11, Chrome, Firefox, Safari) for the MVC web app for instance;
2/ to provide app support for users targeting different environment: prod, hom, dev, etc.;
3/ to provide monitoring tools: performance indicators.
Technical environment : c#, ASP.net (MVC 5), Silverlight (MVVM), MEF, EO.Pdf, JQuery 1.10, CSS 3, Html 5, web api 4.0, twitter bootstrapper 2.3, sql server 2008 R2, T-SQL, WCF, data services, Entity Framework 5.0, log4net, Moq, TFS 2012.
- Directeur R&D
2013 - 2014
Charged to manage and lead R&D projects development.
As a manager, i'm in charge to recruit and validate resources to achieve development goals, assign and scheduling user stories to iterations (scrum).
As a tech leader I was enrolled in development process with the roles of Scrum Master and team member. I supervise and participate in scrum iterations: task decomposition, task evaluation, code review, problem solving, architecture design for Rich Desktop and Internet Applications, 1/3 and 2/3 architecture application, IDE integration by Creating Add-ins and Wizards for Visual Studio 2012.
Technical environment : .Net 4.5, WPF, WCF, MVVM, EntLib, Prism, EF 5, Silverlight, ASP.Net, MVC 4, RIA Services, C#, T4 templates, EnvDTE for VS 2012
Aéroports de Paris
- .Net Developer [Via Ausy]
2012 - 2013
Price fixed Project : implementing an Airport Resources Manager.
• Technical architecture : 2-Tiers Architecture, WCF (dual binding), client WPF ( .Net 4.5, Prism, MVVM, Unity), server (console, service windows, Win server 2008), Oracle 11.2 ( Entity Framework 5.0, Oracle Data Provider).
• Development méthodology: Scrum, TDD (MSTest, Moq, TestRunner, tfspreview).
• Writing technical document : software architecture, technical environment, deployment process, etc.
• Supervising functional documents.
• Technical refering in a team of 4 developers.
Banque de France
- .Net Software Architect
2012 - 2012
• Develop, maintain and package ( Item templates , project templates and vsix) a shared technical framework (based onUnity et EntLib) for building Rich applications such as patterns [ MVC (ASP.NET), MVVM (WPF), SOA, IoC, Composite Applications, muti-threading, data provider, dual binding (for WCF), etc.], and windows docking manager (AvalonDock) for WPF apps.
• Technological foresight related to framework development practices.
• Technical lead assisting project managers at differents steps of the live cycle project developement.
• Monitoring and managing .net projects in Dev, Hom et UAT environment withing Sofact/Bamboo.
• CTI member advising and validating project managers and project team members during the software design phase.
• Tools and software benchmark.
Société Générale - Corporate & Investment Banking (SGCIB)
- .NET Developer
2010 - 2012
• Scrum team member for designing code and bug fixing developer for worldwide (NY/Paris/HK) trading platform and application:
+ Order manager OMS, decommissioning ORION (2M€),
+ Automatic hedging [XHEDGE],
+ Product (Swap, Spot) pricing.
--> Technical environment and tools : .NET 3.5, C#, Rx, WPF, MVVM, Unify/Iron (Unity adapter for IoC), SOA, NHibernate, MySql, Oracle, Design Patterns, Middleware: TIBCO (RV, EMS) and Zulu, multi-threading, real time latency (100ms) and very low latency (1ms).
• Rapid developer team member, providing support 2 and 3 and rapid delivery for commando team around ACSE (excel plugins system and platform). Some example of Excel plugins are:
+ Financial data stream
+ Financial data visualization (ZedGraph)
+ Financial data processing,
--> Technical environment and tools: (.NET 2.0/3.5 : C#, VBA, Winforms, WPF, ZedGraph), multi-threading, real time latency (100ms) and low latency (50-20 ms).
• Other work context elements:
+ Test Driven Development [NUnit, Moq]
+ Code review
+ Multi test environment Dev, Hom and UAT for delivery validation
+ App monitoring
+ Team member of team size from 3 to 8 people belonging to a larger team of 14 people.
+ Reverse engineering in code (Java) and workflow TIBCO (RV, EMS) of existing apps, and designing with Enterprise Architect
CEREMADE & LAMSADE
- PhD candidate (Data mining)
2006 - 2010
Scale up an existing data analysis process called symbolic data analysis. The particularity of this model analysis processing is to compute data basin on complex (non atomic) data.
• The first obstacle was to construct an algebra, called Symbolic Data Analysis (SDA), able to express complex queries for data processing dealing with simple data types (string, int, floats, etc.) and complex one (interval, histogram, probabilistic distribution, vector, etc.). After establishing the state of the art, the relational and nested algebra inspired me to construct a new algebra (SDA) in order to be able to transcript some data mining algorithms into queries in SDA.
• The second obstacle was the implementation of SDA considering technical constraints such as data persistency, data access, data volume, algebraic language for writing queries, query evaluation, data visualization.
A prototype was developed using the CLR of SQL SERVER 2005/2008 (deployment hosts) in order to
+ Store data in SQL SERVER using User-Defined Types (UDT).
+ Compute in SQL SERVER complex aggregates using User-Defined Aggregates (UDA).
+ Compute in SQL SERVER complex operations using User-Defined Functions (UDF).
+ Express algebraic queries using XML syntax against an XSD validator.
+ Implement in SQL SERVER the algebraic operators for query evaluation using C# + Store procedures.
+ A WPF app was developed to send queries for evaluation and display the results (summarized table) in charts (Visifire).
All the implementations were validated with unit tests using NUnit, for the different types, function and queries (data mining algorithms DSTAT, k-means).
• The third and last obstacle was to apply this approach on some real use cases:
+ Enhance the prediction of electricity consumption forecast based on prediction models.
+ Classifying art images of a museum consortium based on RGB histograms.