Engineering principles do not change with the language. Once they are in place and you use AI as leverage, picking a stack becomes an implementation detail — chosen by project context, not team preference.

I build in Rust, Dart / Flutter, Swift, Kotlin, TypeScript, Go, Python, Java — or whatever your business needs. Including technologies that do not exist yet today.

The classic slow path — design, prototype, demo, rewrite for production — is obsolete in the AI era. An AI-augmented workflow produces a prototype that is already architecturally sound, and the same workflow takes it to production.

No rewrite round between prototype and prod. No "MVP code" that gets thrown away. No trade-off between speed and quality.

Generic SaaS has to work for everyone, so it is perfect for no one. You pay the licence fee, you adapt to someone else's IP — and you still do not get what you actually need. (The same problem applies to WordPress, Salesforce, and most of the off-the-shelf software economy.)

AI has made custom software development economically viable for mid-sized companies. Your own application, fitted to your workflow, on every platform you actually use — Android, iOS, macOS, Linux, Windows, web. And maintenance is cheaper than the old "custom is expensive" reputation suggests.

Traditional CI/CD is just the start. I go further: AI generates implementation plans from requirements, AI writes the code, AI generates the tests, AI does first-pass review. The human approves the plan before work starts and the merge request before merge.

Everything in between is automated, auditable, and repeatable. A large team is no longer the prerequisite for high throughput.

Experience across payments and financial platforms, regulated industries, and B2B SaaS. Hybrid environments, multi-tenancy, long-term sustainability.

AI tooling embedded into infrastructure workflows — from infra-as-code reviews to automated capacity planning.

Planner, implementer, reviewer, tester — each agent with a clear role, coordinated through MCP or custom orchestration. The human plays architect and auditor.

Suited to work where a traditional team would need days: codebase exploration, large-scale refactoring, generating and maintaining test suites, content / data pipelines, internal tooling.

Automated tests in 2026 are not written by hand. AI generates unit, integration and E2E from the requirements, keeps them in sync with the code, and is accountable for coverage.

Humans focus on what AI does not yet do well — exploratory testing, domain scenarios, edge cases that only someone who knows the product can spot.

Leadership grounded in trust, autonomy, and growth — the same approach I used at Paysafe and group.one. Now applied in the context of AI-augmented engineering.

The right roles, gating points, and quality bars. Teams ship consistently even through uncertainty. An "AI-first" culture instead of "we will add AI later".