Production AI for real systems Β· RAG Β· Document AI Β· Azure AI Foundry Β· Energy Tech
Playful tools. Serious architecture. Built for the messy real world.
Berlin-based AI Solution Architect building production RAG, Document AI, and enterprise AI systems for energy, real estate, sustainability, and industrial technology.
- π Architected an AI invoice extraction system for Zurich Insurance that reached 94% production accuracy and was recognised as a Digital Top 10 Project 2025 by pom+.
- π’ Built LLM-powered lease contract extraction for a real estate platform managing 370M square metres across 100+ countries.
- β‘ Started in energy infrastructure and high-voltage test equipment, then moved into cloud, AI, and solution architecture - useful when systems touch real assets, regulation, and messy enterprise data.
- π MBA in Energy Management from TU Berlin with an Electrical and Electronics Engineering foundation.
- π Currently sharpening work around AI agents, RAG evaluation, Document AI, Azure AI Foundry, and developer-facing visual tools.
- π¬ Ask me about AI solution architecture, RAG, Document AI, context engineering, distributed systems, energy tech, and technical consulting.
| Signal | What it means |
|---|---|
| 94% production accuracy | Document AI shipped beyond prototype mode |
| Digital Top 10 Project 2025 | Recognised by pom+ / Digital Real Estate Switzerland |
| 370M sqm / 100+ countries | Enterprise platform scale, not demo data |
| Energy + AI background | Domain context for utilities, sustainability, real estate, and industrial systems |
- π€ Production AI: RAG, Document AI, evaluation loops, and AI agent patterns that can be operated responsibly.
- ποΈ Solution Architecture: Turning business problems into systems with clear data flow, governance, and delivery tradeoffs.
- β‘ Energy & Industrial Tech: AI where domain context matters - utilities, sustainability, building tech, and real assets.
- π¨ Useful Visual Tools: Interactive explainers, graphs, and developer tools that make complex systems easier to reason about.
|
|
|
|
|
|
|
|
|
β From niranjanxprt with curiosity, architecture diagrams, and just enough sparkle.



