Martin Kiefer

I am a senior software engineer at Keebo working on query acceleration with data summaries. Before that, I obtained a P.h.D. in computer science at TU Berlin. There, I published at several top-tier venues, coordinated and gave classes, supervised theses, and worked on projects. You’ll find an exhaustive list of my activities below.

My research interests include data management, data summarization, query optimization, and modern hardware.

I received my M.Sc degree at TU Berlin as well. Prior to that, I received my B.Sc degree at Baden-Württemberg Cooperative State University (DHBW Stuttgart) and worked at IBM for three years.



  • I defended my Ph.D. thesis ‘Accelerating Approximate Data Analysis using Parallel Processors’.
  • Our paper ‘Optimistic Data Parallelism for FPGA-Accelerated Sketching’ was accepted for publication in PVLDB and will be presented at VLDB’23 in Vancouver, Canada.


Optimistic Data Parallelism for FPGA-Accelerated Sketching. Martin Kiefer, Ilias Poulakis, Eleni Tzirita Zacharatou, Volker Markl. PVLDB, 16(5): 1113 – 1125, 2023. [Paper | Code]

In the Land of Data Streams where Synopses are Missing, the One Framework to Bring Them All. Rudi Poepsel-Lemaitre, Martin Kiefer, Joscha von Hein, Jorge-Arnulfo Quiané-Ruiz, Volker Markl. PVLDB, 14(10): 1818 – 1831, 2021. [Paper | Poster | Video | Code]

Scotch: Generating FPGA-Accelerators for Sketching at Line Rate. Martin Kiefer, Ilias Poulakis, Sebastian Breß, Volker Markl. PVLDB, 14(3): 281 – 293, 2021. [Paper | Poster | Video | Slides | Code]

Estimating Join Selectivities using Bandwidth-Optimized Kernel Density Models. Martin Kiefer, Max Heimel, Sebastian Breß, Volker Markl. PVLDB, 10(13): 2085-2096, 2017. [Paper | Poster | Slides | Code]

Self-Tuning, GPU-Accelerated Kernel Density Models for Multidimensional Selectivity Estimation. Max Heimel, Martin Kiefer, Volker Markl. SIGMOD, 2015. [Paper | Code]

Demonstrating Transfer-Efficient Sample Maintenance on Graphics Cards. Martin Kiefer, Max Heimel, Volker Markl. EDBT, 2015. [Paper | Poster]


Accelerating Approximate Data Analysis using Parallel Processors. Martin Kiefer. Ph.D. Thesis, TU Berlin, 2023. Summa Cum Laude. [Thesis | Slides].

Investigating GPU-accelerated Kernel Density Estimators for Join Selectivity Estimation. Martin Kiefer. Master Thesis, TU Berlin, 2016. [Thesis Slides]

Evaluation of Possibilities for the Integration of a Large-Scale Computer Architecture’s SMT Functionality in a Virtualization Solution. Martin Kiefer. Bachelor Thesis, Cooperative State Universtiy Baden-Württemberg / IBM, 2013.


Approximate Analysis of Massive Data Streams using Modern Hardware (ADAM). Project Lead. TU Berlin, Huawei, Software Campus, BMBF. 2018-2021.

Autoplanner. Time series analysis on large-scale network data. Storage and infrastructure. TU Berlin, BENOCS, 2018.


US Excursion. Sebastian Bress, Martin Kiefer, Andreas Kunft, Jonas Traub, Volker Markl. Talks and discussions at various companies and universities (Oracle, Microsoft, IBM, Harvard, Stanford, UC Berkeley, others), 2019.

Error-driven sample maintenance in GPU-accelerated Kernel Density Estimation. Martin Kiefer. Invited talk at Data (Co-)Processing on Heterogeneous Hardware (DAPHNE), March 2015. [Slides]

Supervised Theses

Investigating GPU-Acceleration Strategies for Exponential Count Min Sketches. Nils Schubert. Master Thesis, TU Berlin, 2022.

Investigating Stream Summaries for Interactive Data Visualization. Rudi Poepsel Lemaitre. Master Thesis, TU Berlin, 2021.

Investigating an Optimistic Architecture for Data-Parallel FPGA-Accelerated Sketching. Alexander But. Bachelor Thesis, TU Berlin, 2021.

GPU-Acceleration for Count-Min and Fast-AGMS Sketches. Moritz Ruge. Bachelor Thesis, TU Berlin, 2020.

Practical Analysis and Optimization of AGMS and Fast-AGMS Sketch Algorithms. Branimir Pavlov. Bachelor Thesis, TU Berlin, 2019.

Investigating Optimizations for Self-Tuning Kernel Density Estimator Models. Felix Jentsch. Bachelor Thesis, TU Berlin, 2019.

Teaching (at TU Berlin)