Fayrix

GOAL:

To create an inventory optimization system for S7 airlines’ subsidiary maintenance, repair, and operating supply (MRO) organization.

SOURCE DATA:

Thousands of operations of spare parts ordering, delivery, stocking, and periodic maintenance.

SOLUTION: 

The developed system collects and processes data, runs models, and visualizes forecast model results using Python, Node.js, and database engines.

RESULT: 

The performance of the system is robust and stable and helps the logistics department manage their inventory far more effectively.

GOAL: 

To develop software to detect cardiac excitation (hereinafter – activation) using neural networks

SOURCE DATA: 

3 566 763 bipolar and unipolar ECG fragments. ECG fragments were collected from different devices at different stages of technology development. Signals came from different patients – healthy and with pathologies

SOLUTION: 

We solved the problem of detecting an activation in the ECG fragment as a classification task, where 0 is no activation, 1 is activation.

RESULT: 

The developed Deep Learning model detects activations with an accuracy of ≥95%.

THE OBJECTIVES:

• To increase sales

• To improve customer loyalty

THE GOAL:

• to develop an algorithm which would automatically evaluate the relevancy of search results to a search query entered by a user

THE SOURCE DATA:

74 000 search queries on 55 000 products

SOLUTION

•A large amount of source data was provided as text

• We used all varieties of TextMining techniques to identify semantic similarity of words and texts and semantic clusterization

• The resulting model was built as an ensemble of decision trees

RESULT

• The suggested model was 15% more accurate than the standard method for estimating the relevancy of search results to a search query which was based on keyword comparison.

THE OBJECTIVE:

•To increase sales through a higher product page conversion rate

THE GOAL:

to develop an algorithm, which would evaluate the probability of customer clicks on certain products

THE SOURCE DATA:

20 mln products views and clicks were provided

• No data on visitors

SOLUTION

• We built a probability model based on the frequency of products displayed together

• The model showed accuracy of ROCAUC=0.76

RESULT

The web-site conversion increased rate by 14%.

THE OBJECTIVE:

•To increase sales through relevant personal offers.

•To improve loyalty through decreasing irrelevant personal offers.

THE GOAL:

To develop prediction algorithms:

• of total customer spending for the next 30 days for each service/product category

• of individual customer spending in the next month for each service/product category

SOURCE DATA:

•7 mln card transactions by 15K customers, 184 categories of services/products

RESULT

•For Goal 1, an ensemble of models was a solution:time series prediction, linearregression, decision trees

•For Goal 2, we built a model based on statistics for prior periods and extended statistics on various combination of attributes

The solution is processing text materials from key media sources, groups and posts in social networks. It allows to analyze presence of a company and its top managers in media space, identify key topics of reference, monitor information waves. This solution helps to significantly increase efficiency of corporate PR departments.

Key features:

• Continuous (24/7) fully automatic monitoring of media space;
• Multilanguage support;
• Support for major social networks and media;
• Dashboards and reports customization;
• Definition of tone and theme of texts;
• Internal algorithms of machine learning;
• Detection of bots;
• Mobile app for top-managers.

Fayrix developed a mobile app for caring parents to check their child's location using their mobile device.

Key features:

• Location control. The app allows parents to always be able to know, where their child is. The number of children under monitoring is unlimited.
• Control zones. The parent can set up control zones and receive instant notifications about the child's arrival to the zone or departure from it.
• Alarm button. In case the child needs help, all he/she has to do is to start the app and press the button. The app will instantly send the parent a text message with the child's location details.

Inflight Guide App The app uses professional aeronautical information to calculate complicated mathematical and meteorological flight model before the flight. In-flight the app uses sensors as triggers confirming different flight stages and provides passengers with real-time explanations of what happens right then and what, when and why can be expected.

Key features:

• Pre-flight turbulence forecast.
• Exact time flight calculation.
• Current location calculation without use of GPS.
• iOS and Android app.

eHealth Platform

Integrated eHealth platform consisting of mobile application for customers, web interface for doctors and clinics, API for partners. 

The platform gathers and analyses customer data to provide expert recommendations and tips. 

Key features:

- Analytical and recommendation engine;

- Multiplatform support (Android, iOS, web);

- Support of text, voice and video chats;

- Doctor workplace (web);

- API for partners.

Field Force Mobile Workplaces

Mobile workplace is a software for mobile devices (smartphones and tablets) which helps to improve efficiency of remote and field staff performance.

Key features:

- Secure access to corporate data and services;

- Documents management and workflow;

- Task tracking;

- Digital signature;

- Geotracking and route planner;

- Integrated MDM and security tools;

- Support of m-POS terminals.