I'm a Master's student in Artificial Intelligence Engineering at the University of Passau, Germany, building AI systems that are practical, explainable, and production-minded — not just demos.
I work at the intersection of:
🔎 Retrieval-Augmented Generation 🤖 Agentic AI & Tool-Using LLMs
📄 Document Intelligence & OCR 👁️ Computer Vision & Real-Time Inference
⚙️ Backend APIs for AI Products 📊 Applied ML & Data Systems
Previously worked with enterprise data and ML pipelines at IBM, and fraud detection systems at PayPal, gaining hands-on experience with Python, SQL, production ML, and stakeholder-facing reporting.
|
An AI-powered QR portfolio that turns a static resume into an interactive recruiter experience. Recruiters scan a QR code, paste a target role or JD, and instantly see which projects prove candidate-role fit. Highlights: Role-based project matching · JD match mode · Reviewer snapshot · Lead capture dashboard · Follow-up suggestions
|
AI / ML / GenAI
Programming / Backend / Tools
Data / ML Libraries
|
Autonomous AI agent for multi-document reasoning — reads, OCRs, semantically compares, and generates reports across documents. Built for invoice comparison and discrepancy detection.
|
Bilingual (German/English) document Q&A system. Upload PDFs or scans, ask questions in natural language, get document-grounded answers via OCR + semantic retrieval.
|
|
📄 Published research prototype Real-time audio-reactive visualization inspired by the city of Passau. Extracts RMS, tempo, beat, and spectral centroid features and maps them to synchronized procedural visuals — 55-65 FPS on commodity hardware.
|
Real-time body gesture recognition for interactive gaming. MediaPipe Pose landmark extraction with rule-based and ML gesture classification, stabilized with buffering and majority voting.
|
|
Unity-based toolkit for immersive 3D data visualization, inspired by D3.js concepts. Supports interactive 3D exploration for XR/AR visual interfaces.
|
Deterministic digital twin for a 30-aircraft eVTOL fleet: graph-based routing, corridor congestion management, and safety-critical emergency dispatch logic with a live React dashboard.
|
mindmap
root((AI Engineering))
RAG Systems
Document Retrieval
OCR Pipelines
Source Grounding
Evaluation Harnesses
LLM Agents
Tool Use
Multi-Step Reasoning
LangGraph Workflows
Trace Debugging
Computer Vision
Object Detection
Gesture Recognition
Real-Time Inference
Audio & Signal Processing
Spectral Feature Extraction
Real-Time Audio Pipelines
Compression Fundamentals
Backend Systems
FastAPI
Docker
Production Deployment
| 🤖 Reasoning Agents Tools + explainable steps |
📄 RAG Systems Real documents, not demos |
👁️ Vision & Signal Real-time, production-ready |
Building AI systems that are useful, explainable, and production-minded.