It’s one of the new professions emerging in the digital age. It’s also one of the most sought-after profiles in the market, in sectors as varied as finance, industry and healthcare. At the crossroads between the virtual and real worlds, datascientists have the power to transform all types of information into opportunities. An extractor, analyst and modeller of machine, paper and video data, according to Mehdi Brahimi (Lead Data Scientist at Assystem), the datascientist is the Swiss Army Knife of 21st century engineering. Read on for an outline of the job of the future.
Physically, it’s difficult to differentiate a datascientist from any other engineer, but their difference lies first and foremost in their skills. Fluent in at least three computer languages, datascientists are programmers as well as mathematicians, well versed in probability and statistics. And they must also have an in-depth knowledge of the sector they work in, as it’s only by fully understanding businesses and processes that they can identify the data they need to extract in order to create agile maintenance, security and productivity solutions.
“They’re really the Swiss Army Knives of data. They stand at the crossroads between mathematics, modelling, IT and the various engineering professions, whether in the nuclear, transport or any other sector”, explains Mehdi Brahimi, Lead Data Scientist at Assystem. “That’s why it has become a sought-after profession today. And it’s got a great future. Because a datascientist can transform empirical knowledge into mathematical models, facilitate everybody’s work by making the best use of data, extract information from data that isn’t immediately obvious and put in place predictive models.”
A scientist, cyborg and apprentice, all rolled into one
And yet the route to becoming a datascientist is not always a straight road. A geek both at home and work, a fan of the Python language and a volunteer civic tech developer in his spare time, 29-year old Mehdi Brahimi began with a two-year pre-university foundation course followed by three years at the ENSMM engineering school in Besançon in eastern France, which specialises in mechanical and micro-technical engineering. He only really entered the data science universe when working on his PhD thesis.
“When I graduated, I was hired by Alstom to work on signal and sensor data processing. After eight months I decided to start a PhD thesis in conjunction with Alstom. It was a really rewarding experience and even enabled me to file a patent on a diagnostic method for identifying faults in catenary systems using sensor signals and machine learning algorithms, which can be used as the basis for establishing a predictive model for the wear and tear of these overhead lines that supply electricity to trains”, says Mehdi. So it was only once he’d actually started his career that Mehdi’s professional path began to take shape. After finishing his PhD, he worked on algorithms for developing predictive maintenance within an Alstom team, before joining Assystem a year ago. Today, he leads a team of datascientists tasked with bringing data-based solutions to clients.
“Academic education teaches you the basics: how to code, set up automatic learning models, understand big data architectures and so on. But after that it’s about learning the job. That’s when you start gaining experience and feeling an affinity for a certain domain”, Mehdi explains. “At the end of their studies there are people who do very well in areas such as insurance or finance, which have models with data that is already very structured. But when it comes to engineering, they’ll find it more difficult because the data doesn’t have that defined structure. You could say that traditional engineering data science is a mindset, a way of reflecting and thinking. In general, it suits people who have a constant thirst for learning”.
Industry: a challenge for data scientists
In the industrial sphere, data is plentiful but it’s scattered and patchy. “We’re still at the stage of standardising how data is collected and structured. It can be very random and sometimes only available in small quantities. It’s a real challenge, because data scientists are taught that the more data the better”, Mehdi points out.
Yet that doesn’t stop datascience from contributing to the construction of Industry 4.0. Systems engineers, for example, have to read hundreds of documents in order to meet standards and regulatory requirements and to design systems that are reliable and compliant. “Accessing and collecting this data is difficult and time consuming. But thanks to datascience, text mining tools and natural language processing, we can automatically extract information from the available data. This involves using models that enable us, for instance, to identify actual requirements (as opposed to other text) within a document. The key advantages to this are that it saves time and provides assurance that no requirements are omitted, which in turn makes systems design more secure while being able to work more quickly and more effectively”.
The same potential exists in the area of industrial maintenance. Thanks to the deployment of IoT, datascientists can now monitor the behaviour of several machines over time. “This means we can create a digital twin of an asset, which will reproduce the way its system works and predict how it will function taking into account its surrounding environment or frequency of use. And so we will be able to increase the asset’s lifespan, manage it more effectively and flexibly, and ultimately optimise the value chain. It also allows us to carry out reverse engineering by extracting relevant information for designers to enhance their products”.
So wherever there’s data, datascientists can make things more effective, rapid, reliable and profitable. Because data extraction and mining – explaining why a quality fault or a machine failure has happened or optimising engineers’ time – can be used to revise industrial procedures and processes. In so doing, to quote Mehdi, “datascience will create augmented engineers”.
Share :
Something to say ?