Michael Mühlbauer

Data Scientist | Statistics & Machine Learning

Statistician and data scientist with 5 years of experience building predictive models, data pipelines, and analytical tools for large-scale survey, census, and geospatial datasets. PhD research focused on developing novel statistical and machine learning methods, with a track record of translating complex analyses into actionable insights for public-sector stakeholders. Developer and maintainer of an open-source R package and 3 first-authored research papers. Growing interest in AI transformation and strategic adoption of AI in organizations.

Get in Touch View Experience

Research Interests

My work focuses on applying advanced analytics and machine learning to solve real-world problems, with a growing interest in AI transformation and strategic adoption of AI in organizations.

📊

Small Area Estimation

Developing statistical methods for reliable estimates in small geographic areas or subpopulations with limited sample sizes.

🗺️

Spatial Data Analysis

Analyzing geographically referenced data, including density estimation from aggregated spatial information and choropleth mapping.

📈

Kernel Density Estimation

Extending measurement error models for density estimation when precise geocoordinates are unavailable due to privacy constraints.

🌲

Random Forests for Ordinal Data

Applying and adapting machine learning methods, particularly ordinal random forests, for analyzing categorical and linguistic data.

Education

Academic background in statistics, economics, and business.

2021 - 2026

Doctoral Candidate in Statistics (Dr. rer. pol.)

University of Bamberg

Bamberg, Germany

Cumulative dissertation: "Reversing the Coarse: Statistical Methods for Recovering Structure from Aggregated, Sparse, and Discretized Data".

Thesis submitted, currently under review.

2019 - 2021

M.Sc. in European Economic Studies

University of Bamberg

Bamberg, Germany

Advanced studies in economics with focus on quantitative methods and European economic policy.

2015 - 2019

B.A. in Business Administration

OTH Regensburg

Regensburg, Germany

Foundation in business administration and quantitative analysis.

Professional Experience

Building solutions at the intersection of data science, machine learning, and business strategy.

Scientific Research Associate

University of Bamberg

Full-time

08.2021 - Present

Bamberg, Germany

• Built end-to-end data pipelines in R for large-scale statistical analyses (survey, census, and geospatial data)
• Extended an open-source R package (fabOF) with partial dependence plots for explainable ML
• Designed and automated a reproducible analysis pipeline for district-level mobility estimation across 400 German districts
• Delivered an applied data analysis project for the City of Jena (urban safety perception study)
• Designed and taught a graduate-level Python data science course (4 semesters)
• Represented academic staff interests in institute-level governance and decision-making bodies
• Served on the examination board for M.Sc. Survey Statistics and Data Analysis

Student Research Assistant

University of Bamberg — Chair of Statistics and Econometrics

Part-time

01.2020 - 06.2021

Bamberg, Germany

• Contributed to a research project on demographic development

Student Research Assistant

University of Bamberg — Chair of Economic Policy

Part-time

05.2020 - 07.2020

Bamberg, Germany

• Developed course materials for an applied econometrics lecture

Working Student — Strategic Projects

Krones AG

Part-time

09.2018 - 02.2019

Neutraubling, Germany

• Designed and conducted a survey to identify structural issues in after-sales
• Proposed an organizational redesign to improve cross-regional coordination
• Presented findings and recommendations to senior leadership

Innovation Management Intern

Continental AG

Internship

09.2016 - 02.2017

Regensburg, Germany

• Evaluated ideas from a corporate ideation platform for innovation decision-making
• Conducted technology and patent research for senior decision-makers
• Analyzed gamification methods and recommended platform enhancements
• Processed and visualized innovation KPIs for corporate reporting

Technical Expertise

A comprehensive toolkit for turning data into actionable insights and production-ready solutions.

📊

Statistical Methods

Regression & GLMs Mixed-Effects Models Bayesian Methods LASSO / Regularization Spatial Statistics Survey Methodology Small Area Estimation Kernel Density Estimation Monte Carlo Simulation Experimental Design

💻

Programming & Tools

R (advanced) Python (proficient) SQL (basic) LaTeX Typst Git / GitHub Quarto Bash High-Performance Computing (Condor)

🤖

Machine Learning

Random Forests Neural Networks Ensemble Methods Ordinal Classification Predictive Analytics Cross-Validation Feature Engineering Model Validation

🧠

LLM & AI Engineering

Generative AI Prompt Engineering RAG MCP Agentic AI AI Strategy & Transformation

Projects

Open-source tools I build to scratch my own itches as a researcher, and to keep my hands on the engineering side of the work.

Arboretum

Public beta

AI-curated research digests from OpenAlex and arXiv

Download GitHub

A cross-platform desktop app that turns plain-language research interests into weekly Markdown newsletters, delivered to disk or email.

Arboretum sits between passive RSS feeds and active Scholar alerts. You describe what you care about once; the app converts the description into academic search keywords, queries OpenAlex and arXiv for recent papers, asks an LLM (Claude, Gemini, or a local Ollama model) to pick and summarise the ten most relevant, and writes the result as a Markdown newsletter. It can also run a multi-query conflict scan against your own research profile, flagging recent papers that overlap with your work.

Highlights

Plain-language topics → academic search keywords + newsletter title
OpenAlex + arXiv paper retrieval with multi-topic parallelism
Three AI backends: Anthropic Claude, Google Gemini, local Ollama
Conflict scanner: 0–100 overlap score against a research profile
Native scheduling (Windows Task Scheduler / macOS launchd) for unattended runs
HTML email delivery with provider-led SMTP wizard
Secrets in DPAPI / Keychain; never on disk

Rust Tauri v2 React TypeScript Tailwind Anthropic API Gemini Ollama SMTP

Zotero MCP Server

Released

Expose a local Zotero library to AI assistants via the Model Context Protocol

GitHub

A maintained fork of an open-source MCP server that lets Claude, Cursor, and other MCP clients search, cite, and export from your personal Zotero library — including full-text PDF extraction.

Out of the box this server gives AI assistants tool calls for searching items, getting citations and bibliographies, exporting BibTeX files, adding new entries, browsing collections, and pulling full text from PDFs via the local Zotero API. My fork adds the export-BibTeX-to-disk tool, fixes intermittent partial exports, repairs add/update/delete operations against the live API, and rebuilds the PDF-extraction progress reporting around a non-blocking text UI so it works on headless boxes too.

Highlights

Search items, get citations & bibliographies across personal and group libraries
Export BibTeX directly to a .bib file on disk (added in this fork)
Full-text PDF extraction via the local Zotero API with progress reporting
Browse collections; add, update, and delete items from MCP tool calls
Direct PDF file access for faster extraction (added in this fork)
Drop-in for any MCP client — Claude Desktop, Cursor, custom hosts

Python 3.12 MCP Python SDK Zotero Web API PyZotero uv

Publications

Research contributions in statistics, spatial analysis, and machine learning.

Density Estimation from Aggregated Data with Integrated Auxiliary Information: Estimating Population Densities with Geospatial Data

Michael Mühlbauer, Timo Schmid

Submitted 2025 DOI: 10.48550/arXiv.2508.03610

Ordinal random forests in language data analysis

Michael Mühlbauer, Lukas Sönning

Submitted 2025 DOI: 10.31234/osf.io/s4bv2

Estimating Disaggregated Mobility Indicators using Area-Level Models on Survey Data

Michael Mühlbauer

Submitted 2026

Get in Touch

Open to research collaborations, consulting projects, and new opportunities. Let's connect and see what we can build together.