Full Cycle Of Product Development, From A Sketch To Fully Deployed Product

User-generated data is extremely important, sensitive, and valuable. Amount of this data can be huge and the way you store, process and leverage it – is the key to deliver the best services to your customers.

To simplify and flexibly operating user data, our team has implemented a solution that allows you to collect, store and process all the data in one logical structure. This approach made possible to create methods for processing different types of data using one common friendly interface.

• Every user’s action is traced in easily scalable storage
• Operator can compose and start triggers for particular users based on the stored data
• Ability to analyze and weight different user’s actions to produce derived factors
• Unlimited flexibility to provide highly personalized user experience based on custom adjustable complex rules

We have developed an in-house sensor that can collect Heart Rate Variability (HRV) data using only the built-in camera of a mobile phone.

This involved building an algorithm precise enough to detect the basic heart rate needed for HRV analysis. As this requires detecting very low frequency signals, standard FFT based approaches are not sufficient. Therefore, we incorporated a lot of smart enhancements, including new digital filters, statistical methods, and logical algorithms.

The end result has been used in various health care and life coaching applications. One of our licensed users conducted a verification study using a Holter monitor to test the quality of our data. Additionally, our engineering interface allows for the verification and fine-tuning of data processing.

Our in-house sensor provides a simple and convenient solution for monitoring heart rate variability without the need for bulky or inconvenient devices.

According to the WHO, about 1.3 billion people have significant limitations. That's 16% of the population - one in six people.

Our clients understand that increasing accessibility is about expanding the target audience, among other reasons. This is why we achieve proven expertise implementing WCAG 2.0 standard with ААА maximum success criteria.

Our clients' products, including Web and Mobile apps, have been successfully certified by specialized audit companies and partners like Aurora, Microsoft, Google, Limeade, Cigna.

The most of MVPs’ architecture is Monolith. And this is ok, while you’re small. But when you grow, you have to scale-up performance as well. In case of Monolith it cost endless expensive comparing to Service Oriented Architecture (SOA).

Monolithic architecture has a “problem of growth” appearing as a point when Product gets too complicated for further development and even to support an existing state. This happens when you try to bring new features into application which wasn’t written scalable.

The main unavoidable and crucial challenge is to migrate an application from Monolith software architecture to SOA, while keeping production online and heart-beating.

Our team has carried out this kind of work several times, including:

• analysis of business requirements
• analysis of the existing design
• design transition configurations
• isolation of services within the system
• separation and implementation of services

To satisfy clients' request to have a single interface for whole content and product development, eliminate lines between content, system configuration and user data, to provide access for the partners, new state-of-the-art Content Management System has been developed.

CMS serves as a centralized place to create a content, setup and manage all client's applications and services distributed across regions and time zones.

• Physical (infrastructure) and logical separation of PHI sensitive data from other content
• Connection to multiple services DB’s
• Cross-regional multilingual support
• Extended flexible access management system
• Easy integration with external systems
• Easy to use, highly-customizable for complex features UI

Many of our customers are faced with the requirement to comply with laws and regulations concerning the processing of personal data and informing the users about their personal data collection and processing. To meet such customer requirements, we have implemented a consent management system.

Our consent management system includes features for managing texts of documents in different languages, versioning, distributing documents through all channels of consent collection including ability to determine in which region which document should be shown (for instance: the EU’s GDPR, California’s CCPA, or Brazil’s LGPD), consent collection, and consent management.

• central management of document texts
• centralized publishing of documents that should be consented
• centralized storage of user's consents
• checking the availability of the user's personal data based on the documents to which he has given consent

To meet the customer's needs for increasing user engagement, we have implemented a specific recommender system.

Our recommender system is based on two types of traditional approaches in development of recommender systems - content-based approach and user activity analysis approach (a.k.a. collaborative filtering).

Content-based approach focuses on the attributes of items in order to make recommendations. User activity analysis approach focuses on the relationships between users and items.

Individually, each of these approaches has a number of limitations. But the correct combination of these approaches to the recommender systems implementation (the so-called hybrid approach) allows to level the limitations of each of them and build an accurate and scalable recommender system.

As a result, an application that integrated with our recommender system has the following advantages:

• cold start to new user (new user will be associated with similar exiting users and will receive the most suitable content for him)
• cold start to new content item (new content item will be associated with similar existing - content items and will appear to users to whom it is most suitable)
• less suitable content item or content item that is already shown to a given user will appear at the bottom of list
• user will see the content that best suits him

We developed navigation and performance components of a Flight Management System (FMS). Unlike most other new solutions on the market, our solution fully implements all ARINC 424 leg types along with the flight and atmosphere model, which takes into account all relevant performance parameters.

The Key to our solution being resilient is us using a classic proven iterative approach instead of trying to predict all cases, which can be encountered (and then calculating based on interpolation and/or trigonometry), our system decides every moment based on the correct physical model based on current conditions, position and attitude. Every layer is built simulating physical capabilities of a real-world system; hence it cannot command unrealistic maneuvers.

All this is combined with sophisticated display model designed based on guidelines for human factors/pilot interface considerations for avionics design with some new innovative flight path presentation methods.

We are convinced that reliable industrial systems can only be built using conservative deep approach with the focus on real-world application by people in real conditions, and not perceived modern UI convenience (which in our opinion may be dangerous in emergencies). Goal of this kind of solutions is to increase safety.

Our C/C++ development team takes part in the R&D of a proprietary DBMS. To speed up most data processing operations, this database allows to perform calculations not only on the CPU, but also on NVIDIA GPUs.

• Caching & memory management
• GPU & CPU data compressors
• Hash-join operation

With these features, the DBMS can efficiently manage large amounts of data and execute complex queries faster. By using GPUs, the system can offload compute-intensive tasks, such as query processing and data analysis, which reduces query response time and improves overall system performance.

For one of our clients we developed a custom event-driven network blocking service with device integration. This service delivers reliability, ease of maintenance and extension with new devices by providing configuration files and integrating a completely new service.

• SOA based services implemented and integrated into the existing system
• Interaction via RESTful API and Kafka Messages
• Operation with network devices to parse ARP and CAM table entries
• Development of the automated/timed blocking and unblocking procedures
• Preparation of unit tests and documentation with diagrams