Alexey Khursevich once brought the testing infrastructure home. A mobile device farm occupied his home office; an attempt to repair one of its batteries produced a small fire. The subsequent arrangement involved outside specialists, metal shelving and a more suitable environment. There are many ways to learn the importance of maintenance. His had unusually persuasive lighting.
The anecdote belongs to a career spent making software behave beyond the demonstration. Khursevich co-founded Solvd, served as its CEO and CTO, and worked on the infrastructure that lets engineers test their work repeatedly. The impressive word in a software presentation is usually “launch.” His professional interests invite a second question: what happens after everyone has applauded?
When the test lab came home
He began earning as a software developer while still a third-year university student, motivated by financial independence. Freelancing followed, then a business with colleagues drawn partly from his university circle. Early work for West Coast customers meant nights extending past midnight. There was little romance in the timetable.
His own formulation of business success gives repetition its due: calculated risks, together with the discipline to perform familiar work well. That is a useful starting point for understanding a founder whose technical subjects include the machinery behind testing. A routine can be tedious and still deserve to be done properly.
“their ability to do the same routines perfectly”
Alexey Khursevich, on what helps businesses grow, 2022
A project in 2015-2016 supplied another lesson. Working as an engineer and then CTO on a Ukrainian ride-hailing startup, he saw a crowded feature plan delay exposure to users. His later prescription was to release a smaller initial product and improve it through feedback. An ambitious specification can be remarkably good at postponing a useful answer.
A cloud with cables
Khursevich’s engineering record includes MCloud, a mobile device farm, and contributions to Zafira, an open-source instrument for viewing automation results. The two projects address related needs: give tests somewhere to run, then give people a useful account of what happened. A result buried in an unreadable report has rather wasted its entrance.
A Selenium Camp biography places him at Qaprosoft as a vice president, working on Docker and Kubernetes infrastructure. It also records an interest in image recognition for test automation. Those details give texture to the executive biography. His work concerned the environment around the code, including how to deploy it and how to understand its behavior.
His 2020 guide to mobile device farms gets satisfyingly physical. It weighs buying access to a commercial testing cloud against building dedicated infrastructure. The answer depends on scale, budget and the project’s requirements. Ownership brings control and capacity, alongside equipment costs and a continuing obligation to look after the equipment.
He follows that decision down to servers, USB hubs and cables. Device selection should reflect the phones customers actually use. Android and iOS require different arrangements. Network connections, power and the condition of the hardware belong in the calculation. The cloud, viewed from this angle, has an unexpectedly substantial shopping list.
This is an appealing detail in Khursevich’s professional story: the willingness to discuss a cable in public. The language of technology businesses can float a considerable distance above the equipment. Here, the argument returns to what must be purchased, connected and maintained. A founder’s expertise becomes easier to assess when it reaches that level of specificity.
Give the tests room to grow
At Selenium Camp, his presentation on Docker and mobile device farms promised actual production challenges and an account of self-hosted mobile cloud infrastructure. Selenium and Appium, both open-source tools used in testing, appeared in the description. The subject was how to assemble these parts into infrastructure a team could use effectively.
The presentation’s appeal lies in that verb: assemble. An engineering team can have good individual tools and still spend considerable effort getting them to cooperate. Khursevich was discussing the connections between the tools, with Docker helping simplify the deployment of the surrounding services. The difficult part of a collection is often persuading it to become a system.
His TestIstanbul session followed the migration from native deployment through Docker to Kubernetes. It covered scaling Jenkins and Selenium infrastructure, the economics of Amazon spot instances, automatic scaling and a reporting instrument deployed in the cluster. Capacity and cost appeared together. An infrastructure plan has to answer to both.
The practical emphasis continued in a DelEx workshop on building an internal Android and iOS automation farm. Teaching the construction of a system is a different exercise from announcing its virtues. It requires attending to what someone else will need to do, and in what order. Across these appearances, his public subject remained the working arrangement beneath the software.
The distinction matters for the people doing the testing. Their task is to learn something about a product. Every hour spent untangling the environment delays that answer. In that sense, the infrastructure is part of the conversation between a team and its own work. It helps decide how quickly a problem becomes visible, and how easily someone else can investigate it.
September 2022: a larger machine
On September 28, 2022, Solvd announced a majority investment from Siguler Guff. Khursevich was CEO. The business, founded in 2011, then described a team of more than 600 engineers across eight countries. It had grown beyond its early group into an organization delivering work across several regions.
Historical company figures announced in September 2022.
The partnership’s announced purposes were specific: strengthen product capabilities, expand international delivery and add senior sales and engineering expertise in the United States. Drew Guff and Shaun Khubchandani joined the board. For Khursevich, the transaction brought experienced people and relationships into the next phase of the company’s development.
His comments later that year retained an engineer’s respect for conditions. Potential work with the investor’s portfolio companies would still have to make commercial sense; Solvd would need to demonstrate its value and earn the business. A connection could open a conversation. Keeping the customer required something more.
Read alongside the infrastructure work, the investment suggests an interesting continuity. At a different scale, the questions concern capacity, specialist knowledge and reliable execution. That is an interpretation of the career, rather than a claim that a company can be managed like a server. People are considerably less tolerant of being restarted.
Success has to survive the traffic
Khursevich’s September 2023 writing on high-load applications begins well before the traffic arrives. He argues for understanding users, estimating demand and setting explicit performance goals. Latency, capacity and reliability need attention during planning. A popular application can expose decisions that looked perfectly adequate when almost nobody was using it.
He connects those early choices to architecture, development practices and testing. His discussion includes chaos engineering, which deliberately probes how systems respond to failures and disruptions. It is a revealing subject for a founder associated with quality assurance: the unfavorable event belongs in the design conversation while there is still time to prepare for it.
Deployment also gets its share of attention, including infrastructure defined in code, monitoring and recovery arrangements. The concern is sustained operation. His article treats a launch as one event in a continuing engineering process, with decisions about maintenance and scaling made early enough to influence the design.
In February 2024, he turned to governance, risk and compliance spending. The question was whether an organization could tell what its security budget was buying. Generic strategies, overlapping tools and overlooked internal expertise could all distort the answer. His recommendations included prioritizing consequential risks and examining gaps before adding more expenditure.
There is a common sensibility in these two subjects. Define the requirement. Make responsibility legible. Inspect the resources already available. Read together, his arguments suggest a connection between his technical and management work. Expense, activity and a reassuring presentation each need a closer look.
The policy needs a human owner
By July 2024, his published attention included acceptable AI use inside a company. His advice emphasized identifying the tools employees use, defining approved uses and assigning responsibility. Training mattered, as did writing policy in language people could understand. The document needed to work for the employee facing an actual decision.
That moves the discussion from purchasing technology to organizing its use. An approved-tool list, a named contact and a clear account of who owns a problem are modest objects beside the grandeur of an AI announcement. They are also objects a colleague can consult on an ordinary working afternoon.
The executive chapter had changed a month earlier. On June 4, 2024, Solvd announced Adam Gabrault as Khursevich’s successor as CEO, saying the outgoing chief would remain involved with the company. The announcement described a business with more than 800 employees across North America, Latin America and Europe.
Khursevich’s earlier work gives those titles and milestones their human scale. There were phones to connect, test results to make readable and deployment problems to work through. His career comes into focus at that level of detail, where an attractive promise acquires practical obligations. A company can publish a promise in a moment; delivering it means returning to particular tasks, particular equipment and particular people. That is where his engineering story has its weight. The applause ends. Someone still has to make the thing work.