We create neural networks to solve business problems
Automatic content creation or transformation. Allows you to create unique texts, audio files, images, music, poetry, etc.

or Regression.
Suitable for determining age from a photo, making a forecast of stock prices, estimating the value of property, forecasting sales or predicting breakdowns

Применяется для изучения и сортировки большого объёма данных в условиях, когда неизвестны параметры сортировки. Например, сегментация клиентов, потоков входящих писем, влияния событий на биржи и прочее
or object detection. Suitable when you need to find the necessary object: fire, weapons, helmets, masks, cancerous tumors, particles of matter, etc. In general, this is a special case of the Classification, but is often distinguished as a separate type of task
Effective for recognizing which class an object belongs to. For example: what product is on the scales, the type of product on the conveyor, the face at the checkpoint, the emotion in the voice, etc.

Already worked with companies

Top team that has been working together for 5 years

Oksana Mushtak

Technical supervisor. Responsible for developing the project architecture.
Co-author of 14 scientific articles in Scopus, VAK, RSCI and other journals.
Co-author of a patent for the WIZARD recommender system.

Maxim Kudryavtsev

Studio manager. Picks up teams
and has been producing projects since 2008.
Founded the WebTeam.Pro studio.
Clients — Russian Railways, Gazpromneft, VTB Leasing, X5 Retail Group

Evgeny Komotsky

Lead Analyst. Creates machine learning models. Participated in projects for Sberbank and Pyaterochka. Head of the design laboratory "Analysis of social network data" UrFU.

Anton Trunov

Lead programmer. In the team, he is responsible for machine learning, neural networks and integration with the customer's infrastructure. Finalist of the Digital Breakthrough Hackathon 2021

We start projects with analytics and a simple prototype
Finding out the goals of development
We study business processes that need to be improved using a neural network
We evaluate the accuracy of the neural network and calculate the payback
We evaluate the profitability of the launch
We make an inexpensive prototype to see if it is possible to solve these problems using a neural network
We create an initial prototype
Implementing a complete solution
If the prototype is successful, we develop and implement a complete solution. This is the most time-consuming and difficult stage of work
We start the system in combat mode
Launching the project
New projects and articles