Backend — clear, durable, measurable.
I keep complexity intentional.
The rest is implementation.
APIs, consistency boundaries, messaging, and cache. Same discipline from enterprise ops to SaaS MVP.
Recent examples.
HOARYAGE — E-Commerce
Fashion storefront — hero, marquee, sidebar filters, 3-step checkout, admin CMS.
Arter Architecture
Architecture firm demo — admin-backed site, TR/EN, project gallery, contact.
Minik Adımlar — Course Management
Preschool/course center SaaS panel — multi-branch dashboard, students, billing, attendance.
ÜyePanel — Community Platform
Membership community app — company directory, announcements, events, careers, surveys. Mobile + admin.
Architecture & AI notes.
When AI writes the code, unclear specs become the bottleneckWhen AI writes the code, unclear specs become the bottleneck
Agents made implementation cheap; vague requirements now show up as expensive rework in review and QA. What I have been seeing in recent projects.
We generate more code with AI; review capacity is the real constraintWe generate more code with AI; review capacity is the real constraint
Agents made opening PRs easy. In the field, speed often outruns tests and architectural guardrails. How I keep that balance.
Datasette Apps: custom HTML inside DatasetteDatasette Apps: custom HTML inside Datasette
Editorial translation of Simon Willison's post on sandboxed data apps.
Good backend stays invisible. It sits in the right place, avoids performative cleverness. If you take on debt, measure it — and plan repayment.
— Aziz Osmanoğlu, 18.JUN.2026