About
I spent half my career writing code and the other half managing people who write code. What I do now sits at the intersection: close enough to make the technical call, far enough back to defend the business one.
Who is Savaş Doğan?
Savaş Doğan is a freelance software architect and AI integration consultant based in Türkiye. He has been building software since 1992. He served as CIO and VP at an international game studio, ran his own IT firm BTOS for six years, and mentored at Startupbootcamp. He works remotely, in English and Turkish.
- First line of code
- 1992
- In the industry
- 33+ years
- Executive
- CIO / VP · international game studio
- Founded
- BTOS · 6 years
- Mentoring
- Startupbootcamp
- Languages
- Turkish (native) · English (professional)
Career log
The log below is one continuous line from 1992. Two things stand out: the first twenty-one years were field and consulting work, and the period after that alternates between management and hands-on engineering. That alternation is not indecision — it is the foundation of the current work.
- 2026–
Freelance software architect and AI integration consultant
Adding LLM capability to enterprise software, multi-tenant product development, and fractional CTO work. Türkiye and international clients, remote.
Independent
- 2025–26
Software Developer Technical Lead
Technical lead on a multi-tenant CRM for the curtain and home-textile trade. 19 module groups, more than 4,000 automated tests, production on Docker Compose.
Qolay Bilişim
- 2022–25
Mobile & Front-End Developer
Web and mobile interfaces for industrial electronics products, with Angular and Ionic.
ELFATEK Electronic
- 2020–21
Mobile application developer
Flutter mobile development, freelance.
Freelance
- 2017–19
CIO / VP / Head of HR
Technology and people leadership at an international game studio — the period where technical decisions had to be defended alongside budget and hiring.
Beyn Technology
- 2013–19
Founder
BTOS — IT consulting and technology recruitment. Six years working both the client side and the candidate side, which is where I learned why software teams actually fall apart.
BTOS
- 2014–15
Mentor
Technical and product mentoring for early-stage startups.
Startupbootcamp
- 1992–2013
Survey and IT consultant
Twenty-one years of projects spanning field measurement to data processing. The period where software stopped being the point and became the instrument.
Consultant Technologies
What am I working on now?
Three things: LLM integration into enterprise software, on-premise AI architectures where data stays inside the company, and sustainable growth of multi-tenant products. On two of them I also work on my own products, which means I pay the cost of my own advice first.
- Local model measurement. I benchmark open-weight models on my own hardware on a regular cadence, and the results feed directly into client decisions. The series
- Local-first product. A desktop engine where data never leaves the device. Architecture
- Model Context Protocol. Exposing internal systems to AI tools over a standard protocol is the fastest-moving part of this work right now.
How I work
Three rules. One: no development starts before the scope is written down. Two: “done” has a fixed definition — if tests and static checks are not green, the work is not done. Three: if something should not be built, I say so, even at the cost of the engagement.
The most expensive decision in a project is not choosing the wrong technology. It is noticing eighteen months later.
Why this positioning?
The most common failure in AI integration is teams who understand the model but not the system, and whose pilot never clears production. A pilot works in a demo; production demands authorisation, cost control, auditability and an exit path. Those are not AI problems. They are thirty-year-old software problems, and that is where I spent the thirty years.
Last updated: