It’s no secret that the field of Data Science is hot today, there is a lot of work, there are not enough hands, and you can start earning 300 fps even before competitors take courses on self-development and creating their own business from owners of selfies with sports cars. We at X5 Group also actively help young professionals to become mature masters, and we can assure you that, even if 300 fps is a utopia, but 300 fps is quite a reality.
A typical case looks like this: a young intern comes to us, gets a mentor from among senior employees, takes training courses in our X5 Digital Academy, builds up competencies, and in 3-4 months becomes a junior analyst, but in fact a Big Data Analysis Manager, as this position is called in staffing table. And some, and immediately as a Senior Manager, if dragged, sealed and demonstrated.
School of Analysts we are holding the second time, the first launch was for internal employees of X5, the second was launched in October 2020 online for all comers and passed the entrance tests. Graduation took place on June 17th, internships and employment agreements, everyone was in attendance.
Let’s tell a little about the content of the School. It builds on two powerful courses that make up its backbone: Python Programming and Machine Learning. The first is 14 lectures ranging from basics, native data structures and basic syntax through iterators, closures and exceptions to modules, attribute management, and Data Science libraries. Asynchronous programming lessons from Sergey Kabanov, which were not even planned initially, turned out to be a pleasant bonus for the students of the pilot launch.
The second course is classical machine learning: regression and classification, boostings and libraries for them, unsupervised learning and time series, a total of 14 lectures and 14 seminars, homework, Kaggle in-class.
Around the two basic courses are built-in modules on mathematics, statistics and AB tests, SQL and databases, Bigdata and Devops. All modules are about the same size – 20-28 hours of classroom lessons plus homework. In total, a graduate of the School is an almost ready-made analyst who owns the tools of daily work, a stack of technologies and knows the principles of storage and processing of big data. He only needs to work 3-4 months as an intern in order to understand how it is in reality, or he can immediately start working as a junior analyst if he has some working experience gained before entering the School, or in parallel with his studies at it.
It is clear that the content of the School is available to trainees, and they fill the lack of knowledge not only directly at the workplace, but also from teaching materials developed within the School. Consider the range of skills and knowledge that, in our understanding, possesses a junior analyst, or, more sonorously, a Big Data Analysis Manager.
He is displayed in this lovely retro image straight from Excel and says that the junior analyst should be able to code in Python (all of a sudden), write basic SQL queries (: you-don’t-say :), know basic statistics at the p- value, be able to transform numbers into insights, be able to fit predict and know the basic stack of computer technologies: Git, Linux, bash, Docker, Kubernetes and further and further, then at some point there will be a phase transition to middle, and then ad infinitum.
In SQL, we ask at the level of joins, groupbyes and window functions, sometimes we can ask to tell you about indexes, and what they are for. We also ask you to solve a Leetcode easy problem in Python in order to understand how confidently the candidate will cope with everyday problems, whether he knows about the complexity of algorithms, whether he does not forget about edge cases, and indeed whether he writes working code at all.
We want the candidate to understand machine learning at the level of basic learning algorithms with and without a teacher, be able to talk about feature validation and engineering, and know the main types of problems and metrics for them. Everything in the scope of the open course from ODS https://mlcourse.ai/…
Statistics lies at the heart of our AV testing pipelines, which we use to assess the economic effect of the implementation of a particular initiative, therefore knowledge of this area is mandatory in the scope of the maximum likelihood method, maximum posterior probability, methods for testing statistical hypotheses, AV testing itself, and assessment methods. statistical parameters.
The important skill of conducting analytical research is not so easy to evaluate in an interview, here we judge more by the words of the applicant, by his pet-projects, by the way he argues his answers in other sections of the interview.
The last requirement is basic computer literacy, where we include skills in working with Git, bash, basic understanding of software testing, understanding of continuous integration processes.
It seems to us that, having gained theoretical knowledge on the topics of the School of Analysts during the internship, and having worked with real tasks of ad-hoc analysis, product analytics, an intelligent intern can easily apply for the position of a junior analyst at the end of the internship, confidently perform tasks and benefit the company , which we reflect in his payroll.
We at X5 conduct two types of internships: year-round and summer. For the first, we try to recruit graduates of the X5 School of Data Analysts, knowing the guaranteed availability of the knowledge and skills we are interested in, but we also take university students if they can combine study and work for 20-30 hours a week.
But the second one starts from July 1, lasts 2 months and gives an amazing opportunity to work 20-40 hours a week instead of hot summer entertainment on the products and projects of our company. Here are some examples.
The pricing product is the calculation of regular price tags in the Pyaterochka chain of 17 thousand stores in order to achieve the specified business metrics. The product uses graph models, boosts, AB tests, operation research models, curve fitting and much more.
X5 has planning for promotions – optimization of a set of products that are put into a promo. When optimizing, product embeddings (hello matrixfactorization model) are taken into account to exclude substitute products, sales forecast and complex mechanics accounting (buy 2 get 1 free) + promo price optimization. It is safe to say that the product collects all the relevant developments of Big Date in the direction of commerce. We pay special attention to product metrics and quality documentation.
There is also a line of work with external organizations that want to make decisions using X5 data. For example, product suppliers can receive complex reports and, for example, adjust their product line or adjust logistics and manufacturing. Advertising Agencies – Run campaigns targeting the desired segment of X5 customers and measure their results. Financial institutions are interested in improving their own scoring models using our models or finding similar clients in behavior.
As a rule, each team has its own requirements for analysts, but the main goal of all products is to make complete solutions in the interests of business units, so we have SQL development, statistics, machine learning, and various engineering tasks.
Usually trainees are engaged in AV testing, product analytics, writing auxiliary code, working with data, building showcases and dashboards.
You can register for the X5 School of Data Analysts here