ML Engineer | AI Engineer
Based in Hamburg, Germany. Building trustworthy ML systems, LLM agents, intelligent products, and applied research prototypes from research to production.
Currently at Airbus as part of my master's thesis, focused on:
- Trustworthy AI and uncertainty-aware systems
- LLM agents and evaluation workflows
- Applied ML for production environments
Previously worked across analytics, consulting, and software engineering roles in aerospace, telecom, and enterprise tech.
- 6+ years across AI/ML, analytics, and software engineering
- Building trustworthy AI at Airbus with conformal prediction, hallucination reduction, and RAG evaluation
- Published at IEEE TENCON 2019, achieved 23% reduction in worst-case 5G localization error, and built TinyML + LLM product prototypes
- ML/AI: PyTorch, Scikit-learn, Hugging Face, RLHF, Conformal Prediction, Active Learning, RAG
- LLM & Product: Google ADK, LangChain, LangGraph, Next.js, React, Django, Node.js
- Systems & Edge: Python, TypeScript, Docker, Kubernetes, GCP, MATLAB, Kalman Filters, TinyML, TensorFlow Lite, BLE
| Project | What it does | Signal |
|---|---|---|
| MediSprache | German medical dictation to structured clinical summaries using faster-whisper, Google ADK, and a Next.js frontend. |
Docker-first full-stack AI app with live demo |
| MIRA | Generates transit-aware, personalized audio experiences timed to a commuter's exact journey using GenAI, voice synthesis, and route context. | Product-style AI prototype with live demo |
| 5g-positioning | Indoor positioning pipeline using 5G PRS, TDoA estimation, RANSAC outlier rejection, and UKF tracking. | 23% reduction in worst-case localization error |
| hcai | Five interactive Django apps covering AutoML, active learning, explainability, recommender systems, and RLHF. | Broad human-centered ML portfolio in one repo |
| tinyml | SmartSip: a TinyML hydration tracker running on Arduino Nano 33 BLE Sense with on-device gesture classification. |
93.7% validation accuracy, 21.8 KB model |
| cv | Resume generator that turns HTML into PDF using Python, Playwright, and uv. |
Practical developer tooling project |
| Role | Organization | Focus |
|---|---|---|
| Master's Thesis | Airbus | Conformal prediction, hallucination reduction, RAG evaluation |
| Working Student, Data Analytics | Airbus | Uncertainty estimation, calibration, evaluation frameworks |
| Working Student | PwC Deutschland | Analytics and consulting support |
| Senior Systems Engineer | Infosys | Backend Java microservices for T-Mobile retail systems |
| Systems Engineer | Infosys | Frontend engineering, automation tooling, and Angular applications |
- Publication: Feynman Machine: A Cortical Machine Intelligence for Path Detection in Advanced Driver-Assistance Systems - IEEE TENCON 2019
- Certifications: Google Gemini learning badges across application development, data, DevOps, network, security, and cloud architecture
