Technical Product Owner working on AI and voice-platform products at Mercedes-Benz. I have a software-engineering background and enjoy working where product decisions meet APIs, platform architecture, developer experience, and production operations.
My main public project is job-hunter, an open-source Python toolkit for structured, human-in-the-loop job-search workflows.
A Python package and CLI that supports job discovery, fit scoring, resume tailoring, cover-letter drafting, tracking, and PDF generation.
It has two operating modes:
- Agent-assisted workflows for tools such as Claude Code, Codex, and Gemini CLI
- Automated batch processing through configurable LLM APIs
The project is published on PyPI as job-hunter-kit.
I use it to explore practical product questions around:
- Human-in-the-loop AI
- Provider-independent LLM integration
- Configuration-first product design
- Structured evidence and anti-fabrication boundaries
- Agent skills and reusable workflows
- Privacy-aware automation
My other repositories are smaller engineering and learning projects rather than a large AI portfolio. They include:
- A local Python chatbot for learning about LLM runtimes, streaming, and conversation memory
- An iOS language-learning prototype built with Swift
- Python command-line and automotive log-analysis experiments
- Older university and team software projects
Some repositories are forks or academic projects and are kept primarily as a record of my software-engineering development.
- Technical Product Owner for AI and speech-platform products
- Former C++ and Python Software Engineer
- B.Sc. in Computer Science
- M.Sc. Global Engineering Management in progress
- Interested in AI products, platform strategy, developer tools, and conversational systems
I prefer starting with the workflow and the user problem before choosing the technology:
- Understand the existing process and its constraints
- Identify where software or AI can remove meaningful friction
- Define clear system and ownership boundaries
- Build the smallest useful version
- Validate it through real usage before expanding it