Selahaddin Akgünsakgun.com

Istanbul, Türkiye — available for select work

Engineering discipline,
meeting the agentic future.

Engineering leader & systems architect. Leading AI adoption without letting go of the architecture. Builder of Dovilo.

14+ years building production systems at scale — including helping build a FinTech unicorn from the ground up — and eight of them setting the technical direction for other engineers. I set the direction for how AI agents enter a delivery pipeline, and I still own the architecture they run inside.

Get in touch Visit Dovilo
portrait — replace with
assets/img/portrait.jpg
AI-Native Engineering Leader
14+
Years shipping production software
$1.3B
FinTech unicorn built from zero
99.8%
App stability at unicorn scale
Millions
Devices reached by shipped code
01 — About

Leverage, not headcount.

How I work, and what I optimise for.

I build modular, API-first architectures with clear interfaces and agent-native tooling — especially around the Model Context Protocol. My focus is not replacing engineers with AI, but multiplying their output while preserving enterprise-grade quality and long-term maintainability.

Most of my career has been spent on systems that cannot afford to fail: a digital wallet moving millions of transactions, telco commerce platforms, fraud detection pipelines. That work shaped a bias for observability, guardrails and self-verifying systems over speed for its own sake.

I lead cross-functional teams, define requirements from a blank page, and hold the work to a standard rather than a deadline. The details most teams let pass — an inconsistent interface, an unhandled edge case, a metric nobody watches — do not get past me. End to end, a delivery is either complete or it is not shipped.

  • BasedIstanbul, Türkiye — remote-first
  • LanguagesTurkish (native) · English (fluent)
  • EducationBSc Computer Science Engineering, Marmara University
  • Focus nowAgent orchestration · MCP · AI-native delivery systems
02 — Leadership

Leading the work without leaving the work.

Eight years of setting technical direction, and a hand still on the architecture.

Open to Head of AI and AI engineering leadership roles

Teams

People before roadmap

At BWATECH I set the technical direction for the mobile team — architecture, review standards, and the calls that unblock people. At stc pay I grew the Android surface from one engineer into a working team. Mentoring, code review, and standards that let people move without asking permission; influence that does not depend on an org chart.

AI adoption

Strategy with guardrails

Deciding where agents belong in a delivery pipeline and where they do not: tool surfaces, scoped access, review gates, cost and token telemetry. Adoption that survives an audit, not a demo.

Delivery

Predictable under pressure

Requirements from a blank page, alignment across product, UI/UX and stakeholders, and release discipline under regulatory deadlines — judged by what ships and what it costs to run.

03 — Experience

Fourteen years, four companies, one pattern.

Zero-to-one products, then the architecture that keeps them alive.

04 — Building

What I am building now.

An agent-native product, built end to end.

Dovilo

Live

An agent-native, gamified productivity platform: task management, a Pomodoro focus timer and isometric city-building, with a local MCP server that lets Claude Code, Cursor, Claude Desktop, Windsurf and Gemini CLI read tasks, update progress step by step, add notes and manage workflows — with per-project scoping, dry-run approval and real-time telemetry including token usage and cost.

Offline-first architecture with optional cloud sync, signed HMAC-SHA256 webhooks with idempotency and FIFO ordering for n8n, Linear, Slack and LangGraph, five interface languages, and no advertising on any tier.

MCP ServeriOSAndroid WindowsmacOSOffline-firstWebhooks
dovilo.app ↗
  • Role — Founder & sole engineer
  • 2026 — present
  • 4 platforms · 5 languages

Architecture & AI-workflow advisory

Selective engagements: auditing a codebase for agent-readiness, designing MCP tool surfaces and guardrails, or setting up the delivery system a team needs before it adds AI to it. Short, scoped, documented.

Enquire →
  • Engagement — 2 to 6 weeks
  • Remote · CET / GST
05 — Stack

Tools I go deep on.

Not a list of everything I have touched — the things I would stake a delivery date on.

AI & agentic

  • Model Context Protocol (MCP)
  • Agentic workflows & tool use
  • LLM APIs — Claude, GPT
  • Context engineering & reusable skills
  • Evaluation, guardrails, self-verifying systems
  • Claude Code · Cursor · LangChain

Core engineering

  • Kotlin · Android · Jetpack
  • Java · Spring Boot · Microservices
  • API design · REST · Docker · Kubernetes
  • MongoDB · Couchbase · Cassandra · Elasticsearch
  • Angular · TypeScript

Leadership

  • Technical leadership & mentorship
  • System design & architecture review
  • Agile · Scrum · Kanban
  • Observability & performance
  • Extreme ownership
06 — Contact

Reachable, and quick to reply.

Roles, advisory work, or a question about Dovilo — all welcome.

No newsletter, no tracking. Your message goes straight to my inbox.