Scope
Test automation strategy across 5-7 engineering teamsQuality Engineering Leader
Aleksandr Melostnoi - Quality Engineering Leader
I build the quality engineering practice for systems where failure is expensive. Twelve years in software, nine of them in quality: payment platforms handling millions of transactions a day, and now reliability for multi-agent AI systems.
Scale
Payment systems at millions of transactions a dayImpact
Regression feedback cut from hours to minutes
Current
Dasend | quality engineering for a fintech product and for the multi-agent AI systems used to build it
Deepest tenure
Qiwi | 5 years | payments at scale, automation strategy across 5-7 teams, QA leadership
People
Led 5-8 QA specialists | mentored 3-5 engineers into automation | interviewed candidates
Differentiator
Testing Agent Swarms Like Distributed Systems
Most of the conversation about AI quality is about prompts and hallucination rates. I work from a different premise: an agent swarm is a distributed system with non-deterministic nodes, so the dominant defect class lives in task handoff - dropped tasks, duplicates, hangs, delegation loops. Those defects survive prompt tuning and they hide inside runs that CI reports green.
Invariant checking
Offline checking over execution traces, gating every run: task terminality, no double execution, delegation-graph integrity, depth and fan-out limits, token budget.
Deterministic simulation
A full agent swarm replayed from a seed with virtual time and controlled interleaving, so a handoff race reaches the ticket as a seed number, not a story.
Task-lifecycle state machines
Tests for forbidden transitions - double assignment, completion after lease expiry, results from a reassigned worker - plus side-effect idempotency under at-least-once delivery.
Replay corpora
Built from live traces, turning every production failure into a permanent regression that runs at PR level.
Fault injection
Killed and hung workers, network partitions, tail latency, partial success, and workers that report success they did not achieve.
Staged evaluations
Failure attribution to a specific worker, using LLM-as-a-judge and synthetic test generation alongside deterministic checks.
Scope
Quality Engineering Leadership for FinTech, Payments, and AI-Assisted Delivery
Quality owned as a function - strategy, standards, automation architecture and reliability - rather than test writing attached to the end of delivery.
I lead two ways: through architecture and standards, and through the engineers working in them. I still write the code that proves the architecture holds.
My strongest scope is automation strategy across teams: replacing per-team solutions with one shared architecture, integrating checks into CI/CD so they run at merge rather than at release, and defining with the team what "green" actually means.
Scale and domain
Qiwi and Dasend: payment systems handling millions of transactions a day, backend services, APIs and release-critical product workflows in regulated environments.
Leadership
Led 5-8 QA specialists across multiple teams - work allocation, technical direction, performance evaluation and career development. Mentored 3-5 engineers into automation and interviewed candidates.
Working level
Hands-on architecture and standards, with the judgement to decide what blocks a merge and what blocks a release - without brittle process overhead.
Experience
Experience Across Payments, Banking, and Multi-Agent AI
Nine years of quality engineering in environments where a failed release has immediate financial and customer impact.
Dasend
Senior Software Engineer in Test / SDET
Early-stage fintech product company. Quality engineering for the fintech product and for the multi-agent AI systems used to build it: automation coverage, run-level gating, and the measurement layer for agent-assisted delivery. Define with the engineering team what "green" actually means - retry counts, degradation thresholds, statistical significance, flakiness tracked not tolerated - plus coordination metrics as an early regression signal. The resulting tooling is used by several teams.
- Invariant checking over agent execution traces, gating every run.
- Deterministic simulation and replay corpora for reproducible handoff failures.
- Backend and API automation with CI/CD checks running on every merge.
Qiwi
Senior Software Development Engineer in Test
Owned test automation strategy across 5-7 engineering teams building fintech and payment products handling millions of transactions a day, replacing per-team solutions with one shared automation architecture. Set technical direction for automation design across team boundaries and maintained the frameworks those teams built on. Cut regression feedback from hours to minutes by redesigning backend and API automation and integrating checks into CI/CD (GitLab CI, Jenkins, TeamCity). Led 5-8 QA specialists - work allocation, technical direction, performance evaluation and career development - mentored 3-5 engineers into automation, and interviewed candidates. Worked with developers, analysts and product teams from planning onward rather than at the end of the cycle.
- Automation strategy and one shared architecture across 5-7 teams.
- Regression feedback from hours to minutes on systems at millions of transactions a day.
- Led 5-8 QA specialists: mentoring, evaluation, development and hiring.
Home Credit Bank
Senior Software Engineer in Test
Mobile test automation for Android and iOS banking products. Covered mobile application flows, backend and API integrations, and customer-critical scenarios. Worked with QA, development and product teams to improve release confidence for mobile banking deliveries.
- Android and iOS mobile automation.
- Coverage for mobile flows and backend/API integrations.
- Release confidence for mobile banking deliveries.
UCS
Android QA Lead / QA Automation Engineer
Led Android test automation and quality activities for r_keeper, a HoReCa product ecosystem used in restaurant and hospitality operations. Coordinated a group of 4-6 engineers on testing activities and improved automated coverage for business-critical mobile workflows. Worked closely with product and engineering teams to identify risks, validate fixes and improve delivery reliability.
- Coordinated a group of 4-6 engineers on Android quality for r_keeper.
- Quality support for HoReCa product releases.
- Risk-focused collaboration with engineering and product teams.
Expertise
Quality Engineering Leadership, FinTech Depth, and AI Agent Reliability
Where the depth sits: leadership scope, automation architecture, and the reliability work for multi-agent AI.
Leadership and strategy
Quality EngineeringTest Automation StrategyEngineering LeadershipTechnical LeadershipTeam LeadershipMentoringPerformance ManagementTechnical HiringCross-functional LeadershipQuality GatesEngineering Standards
AI agent reliability
AI EvaluationLLM EvaluationAgent ReliabilityMulti-Agent SystemsInvariant CheckingDeterministic Simulation TestingFault InjectionReplay CorporaLLM-as-a-JudgeSynthetic Test Generation
Automation and quality
SDETTest ArchitectureAPI TestingBackend TestingWeb TestingMobile Test AutomationAndroid Test AutomationiOS Test Automation
Stack and tooling
JavaKotlinPythonTypeScriptREST AssuredJUnit 5SeleniumPlaywrightSpring BootDockerKubernetesGCPGitLab CIJenkinsTeamCity
Domain knowledge
FinTechPayment SystemsBanking ProductsRegulated EnvironmentsHoReCa
Credentials
Education, certification, languages, and earlier roles
Twelve years in software, nine of them in quality engineering. The earlier engineering roles stay visible - they are why the quality work reads as engineering, not process.
Education
- Voronezh State Technical University Engineer - Aircraft and Helicopter Engineering
Certification
- Oracle Certified Associate Java SE 8 Programmer - November 2016
Earlier experience
- Quality Assurance Automation Engineer - Antera (Jan 2017 - Jan 2018)
- Software Engineer - T1 Consulting (Jan 2015 - Jan 2017)
- Web Developer - Uz-DaewooAuto (Apr 2014 - Jan 2015)
Languages
- Russian - Native
- English - Limited working proficiency (B1). Strong written English; comfortable with technical discussion and written design work.
- Serbian - Elementary
Contact
I own quality as an engineering function - strategy, standards, automation architecture and reliability.
The work fits engineering organisations where quality is an engineering problem rather than a process one - payments, regulated systems, or products where part of the code is already written by agents. Remote, or Novi Sad / Belgrade.
Currently taking Head of Quality Engineering conversations.