The Bug Stops Here: How One Engineer Is Redefining What Software Quality Actually Means

Boris Vasilev has spent years at the unglamorous end of the software stack — catching what everyone else missed. Here's why that makes him one of the most important people in the room. Software quality assurance doesn't get the conference keynotes. It doesn't attract the VC decks or the breathless TechCrunch coverage. It's the part of engineering that teams talk about after something breaks, and quietly ignore when everything is running fine. Boris Vasilev has spent his entire career in that unglamorous gap. And in doing so, he's arrived at a thesis worth paying attention to: quality isn't a phase at the end of the pipeline. It's an architectural decision made at the beginning. From Saint Petersburg to Florida, One Test Suite at a Time Vasilev's path into QA was a deliberate choice from the start. After completing a Master of Engineering at ITMO University in Saint Petersburg and a focused Software QA program at Yandex Practicum, he went straight to the problem. His first professional role at m2.ru, a high-traffic consumer real estate platform owned by VTB, dropped him into the deep end: TypeScript, Puppeteer, GraphQL, web automation frameworks, and the kind of bug volume that makes you rethink your entire test coverage strategy. He built structured end-to-end coverage before the product shipped, rather than scrambling after it did, and customer-reported bugs fell by 50%. That number matters. Not because it looks good on a resume, but because it represents something harder to measure: the discipline to get ahead of the problem instead of chasing it. The 10% Problem, and How AI Solved It The clearest illustration of Vasilev's approach comes from his current role as Senior Software QA Engineer at UST, a global technology company. When he arrived, the Android Espresso and iOS XCUITest regression suites were running at 10% stability. One in ten runs produced a reliable signal. The rest was noise: flaky tests, false positives, wasted engineering time, delayed releases. His solution was to rethink what was making the tests fail in the first place. Using AI and machine learning techniques, including anomaly detection, ML-assisted test prioritization, and intelligent flakiness analysis, he brought regression stability from 10% to over 90%. Execution time dropped from two hours to forty minutes. This is what applied AI in QA actually looks like. Not a chatbot generating test names, not a dashboard full of green checkmarks. It's a systematic application of machine learning to find the signal in the noise, making the entire CI/CD pipeline faster and more trustworthy as a result. Quality as Infrastructure At Virtu Medical in Miramar, Florida, Vasilev worked across more than eight concurrent projects, improving team velocity by 20% through what he describes as "automation-first initiatives." The phrase is worth unpacking. Most engineering organizations treat automation as a luxury, something you build when you have spare capacity. Vasilev treats it as infrastructure: the thing you build first, because everything downstream depends on it. C#, Selenium, Appium, SpecFlow, Python, Playwright, NodeJS. The stack changed with each project. The principle didn't. He also built self-service QA tooling in NodeJS and PowerShell that let developers run their own test workflows without waiting for a dedicated QA cycle. That shift, from QA as a gate to QA as a capability distributed across the team, is exactly the kind of cultural change that compounds over time. The Peer Recognition Layer Technical output is one thing. Peer recognition is another. Vasilev holds Fellow membership at Hackathon Raptors, a UK-based professional engineering association that grants Fellowship only to individuals who have demonstrated sustained high-level achievement for a minimum of five years. The election process is deliberately rigorous: a peer-review committee, random elector selection, and a requirement of at least four out of five votes to be admitted. It is, by design, hard to get into. In 2026, he was invited to serve as a jury member at Bridge Tech Contest III, an international technology competition spanning AI, blockchain, fintech, and IT, held in Moscow with 213 applications across 14 nomination categories. Jury members at Bridge are selected for domain expertise, not visibility. He also served as a jury member for the Top-40 Digital Experts rating organized by URA.RU, a competitive selection identifying the leading figures driving innovation across Russia's IT industry. Participation as a juror at that level reflects standing earned through years of practical work, not just a professional profile. Beyond competitions, Vasilev participates in peer review for academic and industry publications in the IT sector, including journals indexed in the VAK and RINC (РИНЦ) databases, the principal Russian registries for peer-reviewed scientific research. Evaluating the work of other engineers and researchers before it reaches print is the kind of contribution that rarely gets counted in performance reviews and almost never shows up in a portfolio, but it's exactly how professional standards in a field get maintained. The Stack Is Beside the Point Look at Vasilev's skill set and the first thing you notice is breadth: Python, C#, JavaScript, TypeScript, Kotlin, Swift, PowerShell. Playwright, PyTest, Espresso, XCUITest, Selenium, SpecFlow. REST, gRPC, GraphQL. MS SQL, MySQL, PostgreSQL. BrowserStack, Azure DevOps, Bitrise, GitLab. Load testing, security testing, accessibility testing, distributed systems. The second thing you notice is that none of it is the point. The point is the outcome. Stability from 10% to 90%. Bug volume cut in half. Unit test coverage from 15% to 80%. Velocity up 20%. These aren't vanity metrics. They're the actual measure of whether a software system does what it's supposed to do, reliably, at scale. That's the argument Vasilev makes with his work, even when he's not making it explicitly: quality engineering is about understanding software deeply enough to know where it will break, and building the systems that catch it before your users do. Why This Matters Now The industry is at an inflection point on quality. AI-generated code is shipping faster than teams can review it. Test coverage is increasingly automated, but automation built on bad foundations just fails faster. The demand for engineers who can bring genuine judgment to quality strategy, not just tooling proficiency, is growing at exactly the moment when the supply of such engineers remains scarce. Vasilev sits at that intersection. He knows the frameworks. He's implemented the AI tooling. He's mentored the junior engineers, reviewed the academic papers, judged the competition entries. And he's done it all in service of a single, unfashionable conviction: software that works reliably isn't a nice-to-have. It's the whole job. In a field that loves to celebrate what gets built, that's a perspective worth amplifying. :::info This article was published under HackerNoon’s Business Blogging Program . ::: \