Agentic systems and ten years of analytics turned into production ML.
Data Scientist and AI Engineer with ten years of analytical experience across finance, telecom, insurance, and gaming. I build applied AI systems, currently developing multi-agent forecasting system with graph-based memory at InnoGames with PyMC, Chronos, and Vertex AI on GCP while completing my M.Sc. at Leuphana University. My thesis develops a dual-process agentic forecasting architecture: a fast analyst and a slower auditor that coordinate over structured memory of past forecast errors, evaluated on both a self-hosted open-source stack and a Google Cloud production path.
An agentic forecasting architecture for game revenue.
My M.Sc. thesis develops an agentic forecasting architecture: a fast analyst and a slower auditor that coordinate over structured memory of past forecast errors to improve probabilistic three-month revenue forecasts at InnoGames.
Work in progress through March 2027. Further details once the thesis is submitted.
Ten years in analytics, from telecom M&A research to production forecasting.
My work is closest to problems where forecasting, analytics engineering, and decision support meet. I care about models that can be inspected, monitored, and explained: Bayesian inference, structured memory, and production forecasting workflows rather than one-off notebooks.
I am finishing an M.Sc. in Management & Data Science at Leuphana University as a Deutschlandstipendium scholar while building production ML forecasting systems at InnoGames. From April 2027, I am open to full-time Data Scientist and AI Engineer roles in Germany.
Selected Projects: Agentic AI, knowledge graphs, Bayesian inference
What I work on
I build production forecasting and decision-support systems for teams that need defensible predictions, not just performant models. My current focus is applied LLM and agentic systems for analyst workflows, hierarchical Bayesian inference, and time-series foundation models running on cloud and self-hosted infrastructure. The projects below span all three: knowledge-graph question answering, hierarchical Bayesian customer lifetime value, and a multi-agent forecasting architecture.
Knowledge Graph Question Answering
Team project: SPARQL-based question answering over structured knowledge graphs, enabling natural language queries over RDF knowledge bases.
KVG ML Route Modelling
Team project with a regional transit partner: applied ML to model and predict transportation routes, combining geospatial data with predictive modelling.
Neo4j Graph Analysis of Artist Influence Networks
Graph database solution using Neo4j and Cypher to analyse artists' influence and cluster musical lineages.
Hierarchical Bayesian Pareto/NBD: Replication of Abe (2009)
Implemented and validated the hierarchical Bayesian Pareto/NBD model from Abe (2009) on the canonical CDNOW dataset.
ChefTreff AI Hackathon: Product Detection
Built in 24 hours: a computer vision pipeline for identifying broken or damaged objects from product imagery.
Writing on agentic systems and probabilistic ML
Looking for a data scientist or AI engineer in Germany, Hamburg or Munich, from April 2027?
I bring the most value in forecasting automation, scalable analytics pipelines, and agentic system design, with a focus on open-source and cloud-native architectures.
My work fits six kinds of teams well:
- Insurance and reinsurance analytics. Underwriting segmentation at Absolute Insurance plus current Bayesian forecasting research, applicable to Munich's insurance cluster and beyond.
- Gaming and consumer products. Currently building ML revenue forecasting and analytics infrastructure at InnoGames in Hamburg, including Microsoft Fabric migration and Power BI dashboards.
- AI-native startups and applied AI labs working on LLM systems, agentic architectures, and retrieval. My thesis builds a dual-process multi-agent forecasting system with LangGraph, coordinating a fast analyst and a slower auditor over structured memory of past forecast errors.
- Telecom and network analytics. Eight years at MTS covering market analysis and M&A across Indian, Serbian, Slovenian, and Moldovan markets, relevant to Munich's operator headquarters.
- Global tech firms and platform teams in Munich and Hamburg where forecasting, applied ML, and production analytics scale across products.
- Startups and scale-ups shipping production ML and agentic systems as core product, not pilots.
I'm comfortable in English C1 and German B2. The last decade has been about shipping analytics for international teams across three industries.