Master's Thesis Researcher, Neuro-Symbolic Process Mining
Master’s thesis research in neuro-symbolic process mining (grade 1.3), focused on robust anomaly detection and LLM-assisted analyst workflows.
- Improved anomaly detection in business process mining by up to 47% with TensorFlow-based models.
- Reached 70% of human-level performance while reducing manual effort by 50% via an AI-LLM system using LangChain, Ollama, and MCP.
- Implemented process-mining and machine-learning algorithms in TensorFlow for repeatable experimentation.
- Built Python tools to automatically select relevant SQL tables and anomaly-detector parameters.
- Authored 33 few-shot prompts, JSON schemas, and microservices for structured agent outputs.
Relevant skills: Data Science, LLM development, Process Mining, Python, TensorFlow, MCP, LangChain, Ollama, Pydantic, Pandas, SQL, PostgreSQL, SQLite3, OpenAPI, Hugging Face, Docker microservices.