I am a Senior Software Engineer with a strong interest in Machine Learning, Generative AI and intelligent data-driven systems. I enjoy turning AI models into practical solutions by combining machine learning, software engineering and system architecture.

My recent work focuses on anomaly detection, automated log analysis, Small Language Models, computer vision and AI-based data pipelines. I particularly enjoy evaluating different approaches and optimizing them for real-world constraints such as limited computing resources, latency and heterogeneous data.

I’m interested in the complete path from data to AI-powered applications — from preprocessing and unsupervised clustering to model fine-tuning, inference optimization and integration into production-oriented systems.

Developed a hybrid AI concept for monitoring vehicle applications using machine learning, Generative AI and agent-based workflows.

• Multi-stage pipeline for AI-driven anomaly detection and agentic workflows for root cause analysis
• Template-based clustering reduced 297,136 log lines to 1,578 (-99.5%) for downstream analysis
• Evaluated clustering and classification approaches based on detection performance and runtime
• Fine-tuned a compact SetFit transformer classifier achieving the same accuracy as a significantly larger zero-shot model
• Reduced inference time on a Raspberry Pi 4 from 3.6 s to 0.11 s per log line (~33× faster)
• Evaluated Small Language Models for structured anomaly assessment
• Developed a Generative-AI-based approach for automatic regex generation from example log lines, enabling log-format-independent parsing

Developed a real-time computer vision pipeline during an automotive event using a front-facing vehicle camera and YOLO inside a test vehicle.

• Streamed camera frames from a test vehicle into a YOLO model
• Performed live object detection during driving
• Used Oriented Bounding Boxes (OBB) for object detection
• Exposed live predictions through an API as an interface for vehicle applications
• Enabled participants to build their own applications on top of the AI-based perception service

Experimented with and optimized smaller AI models for resource-constrained and application-specific use cases.

• Prompt engineering and systematic evaluation of Generative AI models
• Fine-tuning of Small Language Models using axolotl and QLoRA
• Model comparison and optimization for latency, resource requirements and task performance
• Integration of Generative AI models into automated analysis pipelines
• Development of agent-based workflows for automated analysis and root-cause investigation

• Machine Learning & Anomaly Detection
• Generative AI & Small Language Models
• Agentic AI & AI-powered automation
• Computer Vision & Object Detection
• Data Pipelines & Automated Log Analysis
• Model Fine-tuning, Evaluation & Optimization
• Edge AI & Embedded Machine Learning
• AI Services & API-based Integration
• Distributed Systems & Microservice Architectures
• Security-by-Design
• Python, C++, Rust & Linux

I’m currently deepening my knowledge of Generative AI, Machine Learning and platform design, with a particular focus on making AI systems practical and efficient.

I’m interested in questions such as how to select and combine different ML approaches, how to optimize models for resource-constrained environments, and how to build flexible pipelines that can turn heterogeneous real-world data into useful insights.

I also enjoy looking beyond the code: communicating technical concepts, mentoring others at hackathons and presenting new technologies on stage.