Evidence-Based Political Reasoning Graph
Designed a multi-stage agentic pipeline to extract and construct verifiable reasoning graphs from raw text, using a Postgres-to-Neo4j flow.
Applied AI Researcher & Data Scientist · Zürich, Switzerland
A background in Data Science and Mathematics, with hands-on expertise in AI research, RAG architectures, and ML engineering.
Designed a multi-stage agentic pipeline to extract and construct verifiable reasoning graphs from raw text, using a Postgres-to-Neo4j flow.
Deployed an automated, containerized pipeline to scrape and summarize market-moving signals using prompt-first workflows.
Designed judgment and refinement pipelines achieving 95.6% hallucination reduction via few-shot/CoT prompting.
Implemented advanced RAG workflows with iterative retrieval and LoRA fine-tuning for model adaptation.
Universität Konstanz
Universität Konstanz
Universität Konstanz
Freie Universität Berlin