The world of Data may seem difficult to access. Here are a few informations to clear thinks up.

/ 01 Data Scientist

Situé au cœur de la chaine de valeur de la donnée, le Data Scientist analyse les data pertinentes à l’aide d’algorithmes élaborés par ses soins et hiérarchise les résultats afin qu’ils soient exploitables par le pouvoir décisionnel. Pour ce faire, il dispose de compétences à la croisée des trois domaines que sont :

1) les mathématiques - régression logistique, modèle bayésien naïf, chaîne de Markov...
2) l’informatique - programmation en Python, connaissance des plateformes cloud et des frameworks Hadoop et Spark...
3) et le business - compréhension des enjeux et des problématiques stratégiques de l’entreprise.

Le Data Scientist doit savoir intervenir sur toutes les étapes du travail : Définition du problème, Collecte et Nettoyage des données, Création de modèles, Implémentation d’algorithmes. Cependant, selon ses appétences et selon les besoins et la maturité de l'organisation à laquelle il appartient, le Data Scientist pourra développer une coloration :

* Business, c'est-à-dire être au contact direct et régulier d’opérationnels métiers pour traduire leurs problématiques en solutions Data et en expliquer les avantages et les limites. Il devra donc avoir de bonnes capacités de communication afin d'adapter son discours à son auditoire. En particulier dans les grands groupes et en l'absence de chef de projet Data dédié, il sera souvent appelé à faire de la vulgarisation pour expliquer au reste de l’entreprise sa démarche et ses conclusions.

* Tech, avec pour objectif d'aider son entreprise à dépasser le stade du POC (Proof Of Concept) en participant activement à la mise en production des modèles et en assurant leur maintenance dans le temps. Son objectif sera alors de délivrer de la valeur immédiate et concrète aux Métiers en leur mettant à disposition des instruments utilisables au quotidien. Il devra pour cela avoir une excellente maitrise de l'environnement IT de production, avec ses contraintes et ses outils.

* Recherche, auquel cas une grande partie de son temps sera dédiée à de la veille sur l'état de l'art et à l'application de concepts et méthodes récents à des problèmes existants. Il interviendra plutôt sur des projets long terme, pour lesquels des retombées économiques immédiates ne sont pas attendues. Dans certains cas, il pourra en outre participer à des conférences spécialisées et publier des papiers de recherche qui feront rayonner son employeur.

/ 02 Data Engineer

The Data Engineer takes care of the daily maintenance of databases and Big Data frameworks. It is also the one who migrates the company's databases and frameworks to the most recent developments. He is responsible for maintaining the Big Data solution developed: he performs tests and evaluations on the structure to make sure that it resists the weight generated by the mass of data used by Data Scientists. He also takes care of the construction of the data pipelines and ensures that they are available for the other Data professions. He has a perfect command of Big Data frameworks such as Hadoop and Spark and is obviously an ace at databases (SQL and noSQL).

/ 03 Machine Learning Engineer

At the crossroads of Data Science and Data Engineering, the role of the Machine Learning Engineer is to optimize and put into production the algorithms developed by the Data Scientist within the infrastructure prepared by the Data Engineer. However, the exact contours of this position have yet to be defined as its emergence is so recent!

/ 04 Data Analyst

The Data Analyst uses statistical and IT tools to organize, synthesize and translate information useful to organizations to guide the decision-making process of decision-makers. He often works on data from a single source that is already known. Downstream in the data processing chain, while collaborating with the Data Scientist on the technical-scientific dimensions, he explores and exploits, extracts and analyses the data by defining relevant KPIs (Key Performance Indicators): he can thus popularise and render the results to decision-makers in a relevant manner and, a fortiori, in a usable format, notably through Data visualisations.

Depending on organizations and their needs, the skills and tools that Data Analysts are asked to master may vary slightly. We can thus distinguish two sub-families:

* The "statistician" Data Analyst who, thanks to his mastery of languages such as R or SAS, will for example be able to carry out customer segmentations (segmentation by purchasing behaviour, Recency-Frequency-Amount segmentation...) and to present them in the form of dashboards (R Shiny).

* The Data Analyst "Business Intelligence" who will be able to explore and cross different databases thanks to SQL or Python data processing libraries (pandas, NumPy), then to extract exploitable KPIs from them thanks to dashboarding tools such as Tableau or Qlik.

/ 05 Data Governance Manager

The Data Governance Project Manager or Data Governance Manager defines and leads the data management program from a business perspective to ensure its quality. Its objective is first of all to raise awareness of the need for data to be collected properly, and then to set up the processes to achieve this, in compliance with compliance rules. In collaboration with IT - which manages the traceability of IS data - and with Data - which enhances the value of the collected data - it determines a data dictionary as well as a common repository and tools.

/ 06 Data Project Manager

Thanks to his or her knowledge of Data issues and problems, as well as business issues (in a sector or company), the Data Project Manager is able to make the link between technical and operational profiles in order to identify relevant Data use cases, prioritise them and monitor their implementation. In doing so, he or she is a major player in the acculturation of an organisation to Data and AI. The project manager generally pilots an aspect of the company's data strategy, such as the management and enrichment of the database or the deployment of Big Data tools. Relational, pedagogical, teamwork and leadership skills are essential.

/ 07 Data Product Manager

Like the Data Project Manager, the Data Product Manager acts as an intermediary between Business and Technical, but with the particularity of working in a product-oriented company. Concretely, his objective is either to develop a 100% Data product, such as a data analysis/IA software, or to develop the Data features of a "conventional" product. This 'product centric' organisation leads to additional prerogatives: in addition to identifying Data use cases, prioritising them and monitoring their implementation, the Data Product Manager defines the Data roadmap for the product and therefore works as closely as possible to the company's long-term strategy. He may have a dedicated team (Data Scientists, AI Software Engineers, Data Engineers, Machine Learning Engineers, Data Analysts...) that he will lead and feed in case of Data use, and with whom he will ensure the maintenance of the developed features and their continuous adequacy with the product strategy.