Stanislav Alekminskii
Backend & Distributed Systems · Hands-on Technical Leadership
I build and evolve backend systems where correctness, scale, and maintainability matter, mostly in Java and Kotlin.
My recent work is event-driven financial transaction processing: system design, production reliability, integration infrastructure that other teams build on, and engineering tooling.
What I work on
Distributed systems
Event-driven processing on Kafka, financial transaction and reporting flows, and the data behind them in MongoDB, PostgreSQL, and Oracle.
Technical leadership
System design, production ownership, code review, and taking ambiguous backend problems from investigation to a working system. Four times I have taken over a departing lead’s responsibilities. Hands-on, not people management.
Engineering leverage
Reusable integration infrastructure, documentation that works as the interface, and experiments with AI-assisted development and review.
Selected work
Scaling financial transaction processing
A financial posting platform for one of Europe’s largest retailers, handling ~4M events a day from 7 sources, with expansion to ~70 planned. In its streaming aggregation model, related accounting lines could expand past the configured stream limit. I designed a replacement built on bounded chunk reads, stateful postings, and size- and time‑based finalization. The implementation is nearing completion.
Integration infrastructure
A reusable adapter library for integrating source systems into the platform: a Spring Boot starter, shared components, source-specific tests, and consumer-contract tests, in one documented multi-module Maven project. Three teams use it now, with a fourth adopting it. The documentation is the first line of support: adopting teams raise almost no questions.
AI-assisted engineering workflows
An experiment in having implementation and review agents work against explicit project rules. A draft pull request gets a Copilot review plus several Claude Code review passes, each with its own focus and a fixed JSON output. A consolidator rates the findings on one scale and returns a verdict to the implementing agent, which applies the fixes. Project conventions live as checkable agent rules instead of prose.
Open source
Hibernate ORM
Fixed invalid SQL generated when an HQL query combines a left join to an unrelated entity with an implicit join in the select clause (HHH-11337). Found on Oracle while building a configurable reporting engine; applied upstream with a regression test.
QuickFIX/J
Connection failures reported to session-state listeners now carry the affected socket address, so a failure can be traced to a host. Found while working on an exchange integration; reviewed with the maintainers and merged for the 3.0.0 release.