Personal experience: preparing for the JetBrains master’s degree at ITMO University and first impressions

Hello everyone! In touch Anton Klochkov, first-year corporate master’s student at JetBrains “Software development“On the basis of ITMO University. I want to tell you how I chose the program, and most importantly – whether my expectations were met.

Northern Eurasia Regional Contests - 2017. Anton in the center of the picture [прим. ред.]
Northern Eurasia Regional Contests – 2017. Anton in the center of the picture [прим. ред.]

A few words about myself

I’m from a small town in the Khabarovsk Territory – Sovetskaya Gavan… In my school years I participated in various Olympiads, but I only won prizes, and I could not go further than the region. I have been writing code since the eighth grade, I have been developing it since the first year of the university.

I graduated from school very successfully: a gold medal and 276 points in the Unified State Exam made it possible to enter ITMO University, which I took advantage of. I chose the bachelor’s program “Programming and Internet technologies”, Where he met many wonderful people and broadened his horizons in the field of software development.

At the end of my second year, I joined 3D4Medical In the role of R&D Engineer, he was engaged in graphics engine optimizations, developed internal tools and implemented features for the company’s application – Complete Anatomy. By the beginning of the fourth year, I changed my focus from development to machine learning and moved to another company – BrainGardenwhere he developed projects related to SLAM (Simultaneous localization and mapping) and computer vision. I was satisfied with the level of salary and the career path was obvious. However, gaps in the field of algorithms and in some areas of mathematics haunted. I understood that a master’s degree is an opportunity to fill these gaps, and decided to choose a suitable program for myself.

Selection and admission

I applied for the corporate JetBrains program after careful consideration and comparison with peers. The criteria were as follows:

  1. There are subjects related to both development and machine learning. I had experience working with deep learning problems, and I wanted to develop in this direction.

  2. The program is based in St. Petersburg – I really like it for its tranquility and accessibility to civilization.

  3. Most of the instructors are either very good at the industry or are working in it right now.

  4. The program is up-to-date. I am sorry to waste my time studying, when the knowledge gained can be used only if you work with technologies from the times of the USSR.

  5. The disciplines are consistent. Knowledge should be supplemented from semester to semester – first you need to go through the fundamental subjects, and then those that are based and build on them.

As a result, my current program, as well as Machine Learning and Data Analysis from the St. Petersburg campus of the Higher School of Economics, came into view. Familiar students of both programs assured that both options are very difficult and interesting. But ITMO University was closer to me in spirit, so I made a choice in its favor.

Unlike most of the guys from my stream, I entered without exams. Even during my undergraduate studies, I was engaged in scientific activities, and by the end of the fourth year I had articles that I attached to the application for the portfolio competition. However, this method might not have brought any result, so in parallel with the submission of documents, I was still preparing for the entrance. For the most part, these were courses on the Stepik platform, which the organizers of the program in Telegram chat… From this year on, they are taken into account as the first part of the entrance test (see. transfer of online courses, and all trajectories of admission can be found at Master’s website). Fortunately, I went through the portfolio competition and waited two months for the start of training, one of which was spent on military training.

A little about study

Each stream has curators. They solve scheduling problems, make learning comfortable and independent of bureaucracy. Taking into account the rather intensive training, they also help to get acquainted with classmates and arrange various extra-curricular activities. In general, you can turn to them on a life issue, they will support and prompt.

Our studies began with a very lamp meeting, at which we ate pizza and got to know each other. Unfortunately, due to the coronavirus, there were two such meetings for the entire semester. Even now, when many universities have returned to full-time education, we continue to study remotely. The fact is that most of the instructors of the program are JetBrains employees, and the company cares about their health and tries to minimize personal contact.

Nevertheless, even at a distance, teachers give material at the highest level. For example, it seems to me that Alexander Vladimirovich Smal (teaches the course “Algorithms and Data Structures”) invented an excellent teaching technique. When a large number of layouts on the board are required, he uses a tablet and streams from it the screen on which he writes. It looks like this:

I also came across a completely new field for myself – functional programming. In this course, we first studied lambda calculus and programming in Haskell. It was interesting, difficult, but we managed it thanks to excellent lectures from Denis Nikolaevich Moskvin and detailed practical lessons.

An experienced commentator will stop me here and say, “Something you have there, everything is too sweet“. And he will be right. As elsewhere, there are problems: I misunderstood something, I didn’t have time for something, I didn’t like something. But all this can be solved. If you do not like how the course is going, you can discuss it with the curators and influence changes in it. I misunderstood something – you write to the teacher, and they will discuss with you the topic of your concern. This is how learning becomes conducive to development.

Research work

In addition to lectures and practices, every semester, students carry out research work (R&D). It should be said that this is one of the most interesting aspects of our Master’s degree program. Students receive topics from companies: JetBrains, Yandex, VKontakte and others, choose the ones they like and work on them under the control of the company’s employees. Specifically, in the first semester, I took on the topic “Generative adversarial audio denoiser” from MynaLabs.

As part of this R&D, I needed to dive into how the audio signal works, what methods of processing it exist, as well as study the nature of noise in the audio signal and train a deep neural network that cleared audio from noise. My supervisor suggested teaching HiFi-GAN – one of the latest approaches to noise cancellation in audio. If you don’t know what a GAN is, then general information can be found here.

For training, I was allocated a separate video card on corporate servers and provided with the necessary data. Throughout the semester, I called my supervisor: we discussed problems, ways to solve them, I described the current results. In general, my R&D was no different from my real work in the company. As a result, I partially managed to train the HiFi-GAN. It seems that I could have achieved better results over a longer period of time, but overall I am satisfied with the work done.

What is the result

The program I’m studying was really difficult. By this word, everyone should understand their own. Non-IT people are likely to find it difficult both in terms of material and in terms of workload. If you have experience in development and knowledge of the basic course of algorithms – do not worry, it will not be easy for you either. The volume of tasks is very large. You should be prepared for the fact that the first semester will be held in the mode of constant painstaking work. You can also do with a strong mathematical background, algorithmic knowledge and many years of experience, but it probably won’t be easier for you: sometimes the problems you need to deal with have a very non-trivial solution.

For me, the most valuable thing in a master’s program is the relationship between teachers and students. The former are interested in us learning something, and try to provide suitable material for the level of students – for example, they adjust the pace of reading so that it is comfortable to listen to lectures and practices. The second – they actively study and discuss the material, plus they maintain a friendly atmosphere even at a distance. If you try to fit my experience into the checklist, you get the following recommendations for those who are thinking about admission:

Assess your motivation. This program is really hard. Therefore, think twice, or even thrice, why you need it. I’m not trying to dissuade, it’s just that I can’t last long on one desire.

Start preparing now. If you took the online test, it is worth preparing for the oral entrance exam. No need to cram, you need to understand. They check not so much for knowledge as for understanding the essence. Of course, you need to know the formula for binomial coefficients and understand how to derive it.

Get ready to quit your job. Yes, as sad as it may sound, it is very difficult to combine work with the first year of master’s degree. There are unique people who managed to work part-time from the second semester, but I have not heard of those who did it in the first semester. In our set, everyone was either taking a long-term vacation or quitting.

In general, my expectations from the magistracy were justified. Initially, I was skeptical about the level of complexity of the program. It seemed to me that it was difficult to build a rocket, but hardly to study. But here I miscalculated. On the other hand, I receive relevant knowledge that will definitely come in handy for me. I am sure because by projecting knowledge on my previous work experience, I find more elegant solutions for different tasks. Now I understand what I did wrong, and how I should. And the time spent will pay off in the future.

What else we have in our blog on Habré:

  • “ITMO Research” – university practice on the topic of UX / UI testing

  • “New Phystech”: a digest of our scientific works and research

  • Clothes with heating elements – made by the startup WARMR

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *