Umutcan
Alanyali
AI/ML Engineering & Consulting
I build AI agents that actually get used. Currently working at Müller Merkle Immobilien, automating research workflows with LangChain and n8n, while finishing a double Master's in Data Engineering and Computer Science.
Building AI that actually gets used.
I got into AI through UX research. Quantitative studies, design systems, behavior analysis. At some point I noticed both need the same core skill. You have to understand how people actually work before you can build something useful for them.
At Müller Merkle Immobilien I noticed that a lot of research workflows were repetitive and saw the automation potential. I proposed AI automations on my own initiative, built the agents in LangChain with a Streamlit frontend, deployed them locally, and the tool ended up actually being used for market research. I also took the proposal to management myself. That's the part I'm most proud of.
What I bring is a mix of technical depth and business sense. Quantitative research at IONOS and 1&1, hands-on AI development with LangChain, n8n and RAG, and a double Master's in Data Engineering and Computer Science.
KI-Portfolio — 3 Live Agents
Drop in any PDF and start chatting. The agent splits the document into chunks, embeds them with OpenAI, finds the most relevant parts for each question and cites its sources. Built with LangChain, ChromaDB and GPT-4o-mini.
Type a company name and get back a full research report. The agent searches the web via Tavily, reads through multiple sources and writes a structured German report on business model, market position and AI potential. PDF export included.
Ten questions across five dimensions: data, processes, team, leadership and infrastructure. The result is a scored report with concrete AI transformation priorities. Animated score reveal and PDF export included.
What I work with.
Where I've worked.
- Built LangChain RAG pipelines for market data analysis on my own initiative
- Built agentic n8n workflows connecting multiple APIs and automating recurring research tasks
- Built a house price prediction model and translated the findings for non-technical decision makers
- Market research and competitive analysis in Python, with pricing recommendations for the team
- Bridged UX standards and AI integration in the product design system
- Designed campaign content for the online shop, documented Figma components and UX guidelines
- Quality assurance in staging environments, cross-team collaboration in the Digital Experience group
- Ran quantitative user studies in Python and turned the data into concrete product recommendations
- Designed digital processes for the ERP system, gathered requirements and collected operational data