Based in Trondheim, Norway. Open to full-time, hybrid and remote opportunities.
I build and evaluate reliable AI-assisted systems at the intersection of multilingual quality, AI safety, agentic workflows, retrieval systems and practical automation.
- AI quality & safety — structured evaluation, factuality review, policy compliance, error taxonomy and human-in-the-loop quality workflows.
- Multilingual systems — Norwegian, Turkish and English language-quality work, localization, terminology and culturally appropriate AI experiences.
- Agentic workflows — AI agents, MCP tool integration, RAG, memory design, workflow orchestration and audit-friendly systems.
- Digital delivery — Python, TypeScript, React, APIs, GitHub, Vercel, documentation, testing and deployment-oriented practice.
| Project | What it demonstrates |
|---|---|
| Multilingual AI services | AI quality, localization and multilingual digital services |
| Professional profile | Recruiter-facing experience, skills and selected evidence |
| Web-ID | React/Vite delivery for multilingual AI-quality and localization services |
| A-Identity-Z | AI agents, multilingual system surfaces and identity-oriented product exploration |
| RAG Project | Modular ingestion, embeddings, vector retrieval, LLM and API workflow design |
| Asana–GitHub Sync | TypeScript automation integration with documented test coverage |
Building practical, testable and explainable AI-assisted workflows for global AI teams, public-sector digitalisation and technology organisations.
- Make claims that can be demonstrated through code, documentation, tests or live delivery.
- Keep user data, credentials and client-sensitive material private by default.
- Prefer small, useful, maintainable systems over unverified platform claims.
- Treat security, accessibility, observability and human review as product requirements.
- Website: multilingual.no
- Portfolio: cv.multilingual.no
- GitHub: @Zekiog
- LinkedIn: zekiogz
- Email: zeki@multilingual.no
Selected repositories represent active work and documented experiments. Learning projects and private research are intentionally separated from this public portfolio.