YURI KORZHENEVSKY RESEARCH AND DEVELOPMENT CENTER

The Korzhenevsky R&D Center is a small collective of experts in Highload, Security and Blockchain

Highload for us mean scenarios where standard architectural solutions are no longer sufficient.
Security is about ensuring that applications, architecture, and team processes work cohesively and protect the system owner’s interests.
Blockchain to us is more about large, sparse systems united by a common protocol, without a single central point of coordination but often with a central point of standardization, similar to many cryptocurrencies.

Currently, our interest lies in SIMPLE (not primitive) solutions for even more complex challenges in competitive areas. Since 2022, we have focused more on our fintech projects, but we are still available for consulting.

Development History of Our Center’s Interests

by founder Yuriy Korzhenevsky

In 2007, I worked on an international Video-On-Demand project, processing hundreds of requests per second on a highly loaded backend. This experience proved that a proper database structure, a deep understanding of MySQL/InnoDB, and simple solutions WORK. Our work led to the creation of a distributed video storage of tens of terabytes across America and Europe, using LustreFS for distributed file storage and lighttpd for delivering 7Gbps to clients.

Later, at Yandex (NASDAQ:YNDX) in the information security department, I developed from scratch a perimeter security monitoring system for the ENTIRE company, integrating security scanners and ensuring real-time resilience. This system, which would have cost over a million dollars in commercial equivalents, is still operational and became the basis for a more functional SoC platform for the company’s information security.

We then explored Erlang, inspired by the reliability of this functional language for the telecom industry, and later added Golang. By 2014, we began early adoption of Golang, finding it promising for backend services development. With a three-person team, we created a distributed transaction storage for a major European bank in two years, combining performance and fast recovery, akin to RAMCloud, and implemented leaderless replication using ePaxos, ensuring linear scalability and real-time data recovery without downtime.

Our Database successfully passed Jepsen — the world’s most challenging test for security and reliability. The test sends data and checks for reliability during continuous artificial disasters on the cluster: delays, node failures, packet losses, clock skews, and drifts. We passed it.

* * *

In 2018, we started assisting blockchain projects with API designs for reliable cryptocurrency storage (custodial services). We adapted cutting-edge technologies and developed a Clickhouse-based IoT analytics and storage system to monitor thousands of printers for a SaaS company. The use of Golang ensured transparent cross-platform compatibility from Raspberry Pi and Windows to MacOS and Linux. In 2020, amidst the COVID pandemic, we delved into low-latency systems, returning to hardware and modern C++.

We were among the first to use Bazel for monorepo builds, focusing on efficient hardware utilization like overclocking processors and memory, and using modern solutions like DDR5, Intel Optane, and ARM. We enjoy writing allocation-free programs in pure C or C++20 with efficient cache-processor data structures and optimizations through SIMD and SVE instructions, favoring XDP and raw UDP, and writing to disk as if data were in memory.

What We Can Offer

We can design, help assess skills and hire the right people, assist investors in evaluating startups, and help founders increase their commercial advantage.

Thanks to our background in information security, we conduct thorough Due Diligence of customers before starting work and thus do not work with scam projects, projects from sanctioned countries, or people with poor reputations.

We do not engage in the turnkey development of mobile applications, e-commerce, and web3 (contracts and ICOs) but can help with complex problems in such projects where we have competence. For example, we have helped a miner corp, optimize protocol and node latency to increase profitability and reduce cloud infrastructure costs.

It’s best to contact us before starting a project. Sometimes, after a year of development, we cannot help if the wrong people have been hired who, despite their best intentions, cannot implement efficient solutions.

We are always happy to meet new people, so if you have a question, a proposal, or just want to get acquainted and exchange experiences, please write to us via email or LinkedIn.

* * *

Contacts

RND.Center Ltd.
Astana, Turkestan 16, Kazakhstan 🇰🇿
Dubai, DWTC Tower, UAE 🇦🇪
[email protected]

© 2015…2023