Today, businesses are increasingly moving away from centralized cloud servers for data storage. These systems can easily become overwhelmed by data streams from equipment, not to mention the high network traffic costs.
Consequently, the modern industry is shifting toward edge computing — processing data directly where it is collected.
Vladyslav Stepanov, a Senior Full Stack Developer and expert at the AITEX IT Association, designs these solutions for major international players, including the manufacturing corporation Jabil and the healthcare company Transcenda.
We spoke with him about how to build fault-tolerant distributed data pipelines and how high-quality engineering choices dictate the viability of IT infrastructure.
Vladyslav, why is the focus shifting toward edge computing right now?
Cloud storage has become too expensive. Sending terabytes of raw data quickly congests communication channels, causing businesses to lose valuable time transferring information back and forth.
In manufacturing or healthcare, delays mean financial losses or safety hazards. Edge computing moves processing closer to the source.
You led the development of the IoT Edge Gateway project at Jabil. What practical problem did this system solve?
Factories run hundreds of automated machines that generate technical data every second.
Our gateway collects these streams locally via industrial protocols like sockets, CFX, AMQP, and REST. The software normalizes the data into a single format, detects equipment anomalies, and sends only critical reports to the central system. This prevents sudden breakdowns.
Scaling this software across factories in Europe, Asia, and the US delivered measurable operational results:
- Manual deployment efforts at facilities were reduced by 40-50%;
- Data transfer latency decreased by ~20%;
- Data integration consistency improved by 25-30%;
- Incident detection time was cut by 30-40%.
What are the main challenges a developer faces when building software for such gateways?
The resource constraints of local hardware. A local factory server isn’t like the virtually infinite space of the cloud. The software must run reliably during network spikes and resume data transmission instantly. We built in rigid buffering and local storage algorithms.
The limited resources of edge devices require an engineer to have a deep understanding of memory management. In an edge architecture, there is no room for redundant code.
Let’s move on to your experience in Digital Health. What are the specifics of working in medicine?
Medical data is both massive and highly confidential, consisting of high-definition images, lab results, and personal health records. Its transmission is regulated by strict international standards like HL7 and FHIR, so you can’t just send a file over a standard protocol.
The system must split datasets into isolated streams, encrypt them, and deliver them to the recipient instantly.
How do you ensure the fault tolerance of such a system?
We utilized the principle of fault isolation. If one node is overloaded or unavailable, the system automatically reroutes the data stream through a backup path.
The data itself isn’t lost; instead, it’s cached at intermediate points. The architecture operates autonomously, without requiring administrator intervention.
How difficult is it to adapt server software to regulatory requirements?
It is the very foundation of design. An architect embeds security rules into the very first lines of code. We configure end-to-end logging, access control systems, and the automatic destruction of temporary data copies. High-quality software protects a company from legal risks and data leaks.
You are on the Board of the AITEX Association. What are your responsibilities within the organization?
I coordinate the community’s technical initiatives and help shape professional standards for backend developers.
The association brings engineers together: we share practical experience, discuss new architectural approaches, and help young specialists adapt to demanding market requirements.
You also served as a jury member for the well-known tech competition, the Code&Create Award 2025. What do you look for first when evaluating other people’s solutions?
I am not interested in the visual exterior; I look at the architectural logic and the cleanliness of the server code. A good solution is evident in how the author handles errors and edge cases.
If an application simply crashes when the database goes down, it’s bad software — no matter how good it looks.
Does judging help you in your primary engineering practice?
Yes, it provides a broad view of the industry, which helps me avoid similar missteps in my own work. Additionally, it keeps me updated on the real technological challenges that businesses face right now.
What is the main quality criterion for modern server software?
Predictability and autonomy. Good software performs its tasks reliably—it handles high loads, caches data locally, and isolates operational failures.
A business shouldn’t even notice the server architecture at work; it should simply ensure the continuity of all processes.
What do you see as the next step in the evolution of distributed data processing systems?
The integration of lightweight analytics models directly into edge devices. Gateways will become even smarter; they won’t just filter data based on predefined rules, but will independently predict anomalies using local context. This will completely transform manufacturing autonomy.
Why Server Software Reliability Determines Business Stability
Modern business is entirely dependent on the stability of its IT infrastructure. Building complex distributed systems and edge gateways demands strict mathematical logic from an engineer, an ability to work under resource constraints, and a readiness to take responsibility for data security at the architectural level. It is these kinds of solutions that define the industry’s technological standard at its current stage of development.
