Software architect based in Istanbul.
A decade of shipping systems that had to outlast the hype around them.
My stack is whatever the problem demands — open-source when possible, managed services when honest.
Architecture is a bet on the future. I try to make the kind that ages well.
I'm not here to sell hours. I'm here to place opinions.
Row-level, schema-level, hybrid — shipped all three. Each has a scar to show for it.
› Done it enough times to have opinions.
Not a fan of the cult. But I know when it's time — and when the monolith is still the right answer.
› Cut when you see the seams, not when the blog post tells you to.
Background jobs that actually finish. Bidirectional streams. Event pipelines. The right tool for the right latency — not the trendy one.
› BullMQ for work. Kafka for events. Pick the wrong one and you'll pay for years.
10-year-old PHP still making money? Respect. Let's give it another 10 years without a rewrite.
› The best legacy is the code that still runs.
Everything a productive team needs — CI/CD, Docker orchestration, deploy automation, observability, monorepo ergonomics. All open-source. Your environment, your rules.
› A team's velocity is set by the day they onboard.
Pre-acquisition, pre-funding, or just pre-regret. I tell you what the codebase won't.
› Every codebase tells the truth if you know how to read it.
Every tool I reach for, watch from a distance, or have buried. Four categories, five concentric zones of intimacy — the closer to the center, the more it's in my daily hands.
// hover any dot · filter by ring · 85 technologies on the map
Job titles change. Stacks change. But the problems tend to rhyme. Here's what I've shipped — repeatedly — across a decade of teams.
// a decade of work · numbers rounded · stories withheld
What changes when I'm there. Left column is what I often walk into. Right column is where we end up.
Hard-won takes from a decade of placing architecture bets. Some controversial. All real. Click any to read the defense.
Microservices are a tax, not a feature.
Managed services are a great product. Your dependency on them isn't.
If your new hire can't docker compose up, you failed.
Documentation is a love letter to your future self.
Kafka is not a database. Stop treating it like one.
The best architecture is the one your team can delete.
Open-source isn't free. It's still usually cheaper.
Multi-tenancy is a business decision, not a technical one.
Observability before scale. Not after.
// nine opinions, each earned the hard way
Every system has four lives: it succeeds, it refuses, it throttles, it degrades. Pick a scenario. Watch where the time goes — and where the decisions hide.
Edge termination, TLS, connection pooling. First line of defense against flood traffic.
# nginx.conf
location /api {
limit_req zone=api burst=20;
proxy_pass http://backend;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Request-ID $request_id;
}Most architects will show you their bookshelf. I'll show you my scars. Five ways I actually get better — none of them involve a chapter summary.
Every time I want to learn a technology deeply, I build a production-ready starter. Auth, queues, CI/CD, Docker, tests — the works. If it survives three real projects, I know it.
The fastest way to learn a system is to break it at 2am in staging. Production teaches you what tutorials can't — the five assumptions nobody wrote down.
If a tool costs $500/month, I want to know why. Usually I can build 80% of it in a weekend. The remaining 20% teaches me whether to buy or keep building.
Not to write the code for me — to argue with me. The best use of a model is as a rubber duck that talks back. Wrong answers clarify my thinking faster than right ones.
Nothing forces understanding like explaining it to someone who doesn't yet. I don't run courses — I run whiteboards. Team pairing, open questions, showing the scars.
I'm suspicious of engineers who only read about the work.
I'm even more suspicious of engineers who only do the work.
The best ones read their own scars.
What I do when I'm not thinking about systems. Or, honestly, still thinking about systems — just different ones.
Guitar. Riffs, rhythm, the patience to practice a phrase until it's clean. Same muscles as clean code.
Motorcycles taught me risk management before any incident postmortem did. Speed only matters when you trust the brakes.
A photographer's eye is a product manager's eye: framing, hierarchy, what you leave out of the frame.
Found something broken? A decision that puzzles you? A system you can't untangle? A contrarian take to argue with me about?
cihantaylan@cihantaylan.comReplies come when they come. I read everything.