Software Architecture: Why We Decided It This Way

Alexey Grishin joined Tinkoff as a lead developer, spent years on the enterprise service bus, and then became an architect. Seeking a domain with a visible pain, he joined Tinkoff Business. Moving to several teams required a repeatable process instead of personal design work.

Teams adopted C4 diagrams, a repository, and ADRs preserving context, alternatives, and consequences. The architect first helped and demonstrated value, then transferred work to engineers with peer review. Iterative decisions could evolve with the product instead of freezing one design forever.

EventStorming supports discovery and complex-domain refactoring. A shared event map reveals business language, conflicting mental models, hidden dependencies, and change boundaries. It works best in a difficult domain; a simple problem or technical platform may not justify a large session.

Changing the process needs management support and evidence from a limited trial. An ADR restores context and explains a past decision. Architecture need not lead into management: interest in design, a real problem, feedback, reflection, and knowledge from adjacent fields matter more than a prescribed ladder.