Michael Zinsmeister Portrait
Hello, I'm

Michael Zinsmeister

PhD Student @ TUM Chair for Data Science and Engineering
Database Systems Researcher
Co-Founder & Board Member of TUMuchData

Personal website of Michael Zinsmeister, PhD Student in Database Systems at TUM

Get In Touch

About Me

I am a PhD student at the Technical University of Munich (TUM) Chair for Data Science and Engineering, advised by Prof. Thomas Neumann. My research focuses on high-performance buffer management for database systems that must handle datasets larger than main memory while leveraging modern flash and NVMe storage. Previously, I worked as a dual student and working student (Werkstudent) at NOVENTI Health SE, gaining hands-on experience with production systems, Docker, and large-scale migrations.

Current Work

PhD Student at the TUM Chair for Data Science and Engineering (advised by Prof. Thomas Neumann). Actively working on high-performance database systems (Umbra).

Community

Co-Founder & Board Member of TUMuchData, organizing database reading groups, guest talks, and coding challenges for systems students at TUM.

Research & Publications

My research is centered around systems-level database performance, specifically in making high-performance buffer management techniques practical and portable. I am an active contributor to the Umbra database project at TUM, building systems optimized for modern hardware.

Featured Publication

Predictive Translation: High-Performance Buffer Management Without the Trade-Offs

Michael Zinsmeister, Lam-Duy Nguyen, Viktor Leis, Thomas Neumann
SIGMOD 2026 (Proceedings of the ACM on Management of Data)

Overview: Traditional database buffer managers rely on simple hash tables—robust and portable but slow. Newer designs are fast but complex and less portable. Predictive Translation delivers state-of-the-art performance using deterministic page placement and CPU latency hiding, while remaining portable and easy to integrate.

Read Paper GitHub

Maintaining a Sample Under Concurrent Updates

2024
Michael Zinsmeister
Master's Thesis, Technical University of Munich (TUM)

Master's thesis focusing on the design and implementation of database sampling algorithms that maintain accurate samples under highly concurrent update streams.

Read Paper

Introduction of an object storage solution into an existing software system

2022
Michael Zinsmeister
Bachelor's Thesis, Hochschule München | Munich, Germany

Designed and integrated an S3-compatible Object Storage solution into an existing pharmacy billing system during my dual studies.

Experience

PhD Student / Doktorand

June 2024 – Present

Technical University of Munich (TUM) • Garching, Germany

  • Researching high-performance buffer management inside the Umbra database project.
  • Published first-author paper at SIGMOD 2026.
  • Assisting in teaching database systems and transaction systems courses.

Co-Founder & Board Member (Infrastructure & Finance)

Nov 2023 – Present

TUMuchData (Munich Database Club) • Munich, Germany

  • Co-founded the official student club at TUM dedicated to database systems.
  • Organizing technical talks, reading groups, coding challenges, and study trips.
  • Authored a technical post on building a safe remote code execution system for coding challenges using Rust and Docker.

Dual Student & Working Student (Werkstudent)

Sept 2018 – April 2024

NOVENTI Health SE • Munich, Germany

During my practical time at NOVENTI i did work on several different tasks within the software system of the pharmacy factoring and clearing systems department, among them:

  • Introducing Docker into the system in my practical semester
  • Introducing a S3-compatible object storage solution into some existing parts of the system in the bachelors thesis
  • Preparing big parts of the migration to Java 11
  • Development of a server side application for visualizing electronic prescriptions

Featured Projects

OxidSQL

A toy SQL database written in Rust. Excellent demonstration of database internals knowledge and modern systems programming practices.

Rust SQL Parser Database Engine
GitHub

PrediCache

C++ prototype implementation of the SIGMOD 2026 Predictive Translation buffer manager. Includes build and execution harnesses for TPC-C and YCSB workloads.

C++ Buffer Manager TPC-C / YCSB
GitHub

FerrisJIT

Rust research experimentation with JIT (Just-In-Time) compilation utilizing inkwell (LLVM wrappers) and copy-and-patch strategies.

Rust LLVM / Inkwell JIT Compilation
GitHub

Teaching & Community

Academic Teaching

I co-instruct and assist in database courses at the Technical University of Munich:

Winter 2024/25
Programming Database Web Applications
TUM
Summer 2025
Transaction Systems
TUM
Summer 2026
Transaction Systems
TUM

Honors & Scholarships

Deutschlandstipendium (Multiple times)
National scholarship for high-achieving students demonstrating academic excellence and social commitment.
Check24 GenDev Stipendium
Scholarship awarded to exceptional computer science students.

Technical Expertise

Core Systems & Databases

Database Internals Buffer Management Query Execution In-Memory Systems Flash/NVMe Storage

Programming Languages

Rust C++ Java Kotlin Go SQL Sass/CSS

Infrastructure & Dev

Docker Linux Systems Object Storage (S3) LLVM / JIT Compilers Git CI/CD

Get In Touch

Contact Information

Feel free to reach out for research collaboration, questions about Predictive Translation, or database systems inquiries.

Academic Email michael.zinsmeister@tum.de
Office Phone +49 (0)89 / 289-17259
Office Address Room 02.11.042, Boltzmannstr. 3, 85748 Garching

Connect Online

GitHub View profile →
LinkedIn View profile →
Google Scholar View profile →
TUM Profile View profile →