How to Squeeze All the Extras Out of an IT Master's Degree and Why 1 Year is Better Than Two

As you probably know, we have already launched many interesting IT Master's degree programs at MAI. We have a program for product managers, where we improve our skills in working with an IT product, many programs for ML specialists and machine learning applications for various purposes, backend specialists, information security specialists and other members of the IT community will find their programs.

At the same time, we have previously somehow ignored such an interesting profession as an ML engineer. About creating a training program for ML engineers “Big Data and Machine Learning“, and even in one year, and this article will discuss it.

However, we are not some kind of “shop” that promises everything at once and delivers it beautiful papers diplomas of their own sample. And therefore they must prepare programs in accordance with the requirements of the State and the Federal Educational Standards FSES 3+, 4+, 5+, SUOS no worse than FSES… with ROPs, FOSs and others abbreviations that Chat GPT 4o can do documents that are constantly changing and contain sometimes useless very important for IT requirements for studying various competencies, in the spirit of:

# ФГОС 01.04.02 Прикладная математика и информатика
competention = {'UK-4': 'Способен применять современные коммуникативные технологии, в том числе на иностранном(ых) языке(ах), для академического и профессионального взаимодействия',
                'UK-5': 'Способен анализировать и учитывать разнообразие культур в процессе межкультурного взаимодействия',
                'UK-6': 'Способен определять и реализовывать приоритеты собственной деятельности и способы ее совершенствования на основе самооценки'}

We are generally for high self-esteem of our students, but we don’t want to spend time on developing this important part of life. There is a whole army of coaches and personal growth specialists for this.

As a result, the curriculum includes disciplines common to all areas – Cultural Studies, Foreign Language, Philosophy. In general, this may be good for regular Master's programs, but it is an extra burden and, to be honest, all IT specialists are perfectly able to read GitHub repositories and documentation on various libs and softwares software products, and many who enter the master's program already have a language level of B2 and higher communicating with Indians on discord with colleagues from different countries. And the applicants already have everything in order with intercultural interaction.

And so in 2024, at MAI, as well as in other 6 leading universities allowed to form specialized higher education programs based on our own requirements! Hurray! We have finally been trusted to assemble programs ourselves and determine the training standards that IT companies and those who want to improve certain skills ask of us.

We leave only what is necessary

So, first of all, we carry out a stage of deep squeezing out all the unnecessary things from the program – this is Foreign Language, extra hours of R&D and Practical (it’s not like you’re practicing for two years), as well as Cultural Studies, Philosophy, Psychology and Intercultural Communications.

Let's decide who we want to get after a year of training! Based on the results of our analysis of vacancies, ML engineers are now in high demand, and with skills in the field of big data. Naturally, the first thing we did was talk to specialists from companies such as Sber, Cloud, Yandex, T1 and many others, where our graduates already work.

We collect and structure the main requests that were voiced for the ML engineer:

1. Python Programming for ML Algorithm Development

2. Deep understanding of ML algorithms (mathematics, programming)

3. Big data processing (cleaning, transformation and visualization)

4. Working with BigData solutions (Apache Hadoop, Spark or Kafka)

5. Ability to evaluate external and internal metrics of ML models (i.e. not only internal ones – MSE, F1, etc., but also the running time of models on hardware, etc.)

6. DevOps and MLOps (both)

7. Rapid prototyping of ML products – so you can feel the results of your work

8. Communication and teamwork (where would we be without Soft Skills)

Designing a program

Based on this, we began to build the program. First of all, we formed the main disciplines and defined the connections between them. We thought about what should be at the beginning, and what you need to know to move on to the next blocks of training.

Here's what we managed to include for training in the fall:

1. Python: Advanced level with all the necessary libraries for work

2. Databases: Advanced Level – About Large and Very Large Databases and How It All Works

3. Mathematics for DataScience – yes, yes, mathematics and more mathematics – without it you won't understand how it all works

4. Machine learning – here about models and their metrics

The profile block was supplemented with a section on the development of related skills and software:

1. Your choice: Product design and prototyping of AI products for those who want to dive into the product part or Frontend development for artificial intelligence systems, for those who are more interested in how it all works “under the hood”

2. Fundamental and promising concepts of artificial intelligence to understand where we are all going and how our world will change when we create strong AI

3. Agile development methodologies – developing teamwork skills using different Agile technologies.

Well, right from the beginning of the training, project practice with a curator starts – practice, practice and more practice. And so on until the New Year!

And in the spring those who will survive Students who are already advanced in ML will study more complex courses for deep immersion:

1. Deep learning and reinforcement learning

2. Deep Generative Models

3. Collecting, generating and labeling data for machine learning (not by hand, of course)

4. Application Containerization and Orchestration

Plus, this half-year will also offer a choice in terms of specialization. Our partners have identified two major areas for specialization – LLM or large language models and data visualization issues, for various applications. Based on this, we have the following block to choose from:

1. Deep Learning and Natural Language Processing (Large Language Models) or Computer Graphics and Design Based on Artificial Intelligence

2. Data-Driven Solution Development and Predictive Analytics or Business Intelligence and BI for Big Data

3. Design and Development of Recommender Systems / Data Analysis and Visualization in Intelligent Systems

Of course, there will also be a lot of practice in this semester. And all this ends with the defense of the diploma project.

To make your studies comfortable, we have our own IT floor with convenient laboratories and a bunch of project teams. So come to us!

This is what the training program turned out to be.

How to proceed

Stages of admission to the specialized higher education program – Master's degree “Big Data and Machine Learning”:

1. Submit your application before September 5 via your MAI applicant account. You must already have a bachelor's or specialist's degree.

2. Take the entrance exam, which will take place on September 10.

3. Conclude and pay for the contract. The cost of training is 148,650 rubles per semester and 297,300 rubles for the entire period of study.

4. Expect the enrollment order on September 28.

5. The training starts on October 1st.

If you like it, you can fly to our telegram chat – we are waiting for you!