I design and ship AI agent systems, automation workflows, and multilingual digital infrastructure — with a focus on production reliability, observable execution, and human-in-the-loop control.
Core areas:
├── AI Agent Orchestration → durable, auditable, human-gated agentic systems
├── Automation & Tooling → N8N, MCP servers, FastAPI, TypeScript pipelines
├── Multilingual Systems → localization engines, translation QA, 3+ languages
├── Self-Hosted Infrastructure → Oracle Cloud, Cloudflare, Supabase, containerized deploys
└── Developer Tooling → open-source templates, CI/CD, secure-by-default repos
| Project | Status | Description |
|---|---|---|
| Vibe-Coding Control Plane | 🔴 Active | AI-assisted coding + agent-builder control panel |
| Multilingual Workflow Engine | 🔴 Active | Localization and content adaptation with AI |
| Self-Hosted AI Infrastructure | 🟡 Ongoing | Apple Silicon + Linux local model deployment |
| Agent Orchestration Templates | 🟡 Ongoing | Production-ready agent patterns + AGENTS.md standard |
Production repos live under @Zekk-s; experiments and forks live here at @Zekiog.
| Repository | Description |
|---|---|
| ai-localization-demo | Real-time AI-powered translation and localization demo |
| agent-builder | Visual agent orchestration and control panel |
| multilingual-webapp-starter | Next.js + i18n boilerplate for multilingual products |
| infra-blueprints | Self-hosted cloud infrastructure templates |
| Repository | Description |
|---|---|
| A-Identity-Z | AI agent identity & runtime context |
| hermes-legal | Multilingual legal/AI tooling |
| rag-project | Retrieval-augmented agent runtime |
| memory-core-mcp | Persistent memory layer for agents |
🇹🇷 Turkish (native) · 🇳🇴 Norwegian (fluent) · 🇬🇧 English (fluent)
Plus working knowledge of: 🇸🇪 Swedish · 🇩🇰 Danish · 🇦🇪 Arabic · 🇩🇪 German
Multilingual background is a core capability — not just a personal fact. I build systems that work across languages, cultures, and locales.
AI & Agents
Infrastructure
Frontend
delivery:
- PR-first: agents and humans both work through pull requests
- main is always deployable
- every change has a test and a reason
security:
- fine-grained tokens only — no god-mode PATs
- secrets in secret managers — never in repos
- signed commits where tooling permits
- quarterly access audit: tokens, collaborators, exposure
agent_policy:
- agents propose via PR, humans approve for production
- AGENTS.md in flagship public repos (in progress)
- no direct push to main by automation
- observable execution with audit trail
infrastructure:
- self-hosted first, cloud-native second
- reproducible deployments (Docker + IaC)
- environment-gated production releases
