I am a developer relations advocate at DuckDB Labs, the company behind the DuckDB in-process analytical database management system, based in Amsterdam, the Netherlands.

Previously, I was a post-doctoral researcher at the Database Architectures group of CWI, where DuckDB was originally conceived. My research focused on creating standard graph processing benchmarks for the Linked Data Benchmark Council.

GitHub – Twitter – LinkedIn – Stack Overflow – Hacker News – DBLP

News

2024
- Mar: I gave a talk at my at the Simonyi Conference, organized by my old alma mater, the Budapest University of Technology and Economics.
- Feb: I was a co-organizer of DuckCon #4.
2023
- Dec: I gave a talk at OSA Con 2023 titled An Overview of DuckDB. See the recording on YouTube.
- Nov: I gave a talk at Øredev 2023 titled DuckDB: Harnessing in-process analytics for data science and beyond.
- Aug: The VLDB organizers granted us the Best Experiment, Analysis, & Benchmark Paper award for the paper “The LDBC Social Network Benchmark: Business Intelligence Workload”, co-authored by me, Jack Waudby (Newcastle University), Ben Steer (Pometry), David Szakallas (LDBC), Altan Birler (TUM), Mingxi Wu (TigerGraph), Yuchen Zhang (TigerGraph), and Peter Boncz (CWI). For an overview of this work, see the slides of my presentation.
- July: Four papers I co-authored are going to be presented at VLDB 2023 and its satellite events.
- July: I started working at DuckDB Labs as a developer relations advocate.
- June: I co-organized the 16th LDBC TUC meeting co-located with SIGMOD 2023 in Seattle.
- June: The DuckPGQ demo paper, authored by Daniel ten Wolde, Peter Boncz, and myself, was accepted at VLDB 2023.
- March: I am the web chair of CIDR 2024.
- Feb: I gave two talks at FOSDEM 2023:
  - Graph Systems and Algorithms devroom (Graph Systems and Algorithms developer room), co-presented with David Püroja
  - The LDBC benchmark suite (HPC developer room), 10-minute lightning talk
- Jan: David Püroja defended his MSc thesis on the LDBC SNB Interactive v2 workload.
- Jan: I served as a local arrangement chair of CIDR 2023.
- Jan: Our paper DuckPGQ: Efficient Property Graph Queries in an analytical RDBMS was presented at CIDR.
2022
- Dec: Our paper The LDBC Social Network Benchmark: Business Intelligence Workload appeared at PVLDB.
- Nov: TigerGraph and AMD conducted the first successful audit of the LDBC SNB BI workoad.
- Sep–Nov: I was a teaching assistant of the Large-Scale Data Engineering course at VU Amsterdam.
- Aug: Danïel ten Wolde defended his MSc thesis on adding SQL/PGQ in DuckDB and decided to pursue a PhD in our group!
- Jun: I organized the 15th LDBC Technical User Community meeting co-located with SIGMOD 2022 in Philadelphia. I was the session chair of SIGMOD’s “Benchmarking and Performance Expriments” research session.
- May: I was the local chair for the (remote) GrAPL 2022 workshop and acted as session chair.
- Apr: I was a reviewer for GRADES-NDA 2022.
- Jan: I was the Zoom chair of the CIDR 2022 conference and co-authored an experience report on running a hybrid conference.
- Jan: I was a reviewer for TheWebConf 2022’s industry track.
2021
- Sep–Nov: I was a teaching assistant of the Large-Scale Data Engineering course at VU Amsterdam.
- Oct: I appeared on the Code for Thought podcast with Benjamin Steer.
- Sep: Our paper A cross-technology benchmark for incremental graph queries was accepted at the Software and Systems Modeling journal.
- Aug: I organized the 14th LDBC TUC meeting co-located with VLDB 2021.
- Apr: Our paper LSQB: A Large-Scale Subgraph Query Benchmark was accepted at GRADES-NDA 2021. The microbenchmark described in this paper was adopted by a number of GDBMS vendors as a performance regression test.
- Mar–Apr: I was a program committee member of the MEDI 2021 conference and the GRADES-NDA 2021 workshop.
- Jan–Mar: I served as an external reviewer for VLDB 2021.
- Feb: I was a program committee member of the EDBT 2021 conference.
- Jan: I served as the local arrangements chair of the online CIDR 2021 conference.
2020
- Dec: I co-authored a preprint documenting our lessons learnt during the implementation of the SIGMOD 2014 Programming Contest.
- Sep: I presented our paper A GraphBLAS solution to the SIGMOD 2014 Programming Contest using multi-source BFS at HPEC 2020. Slide deck
- Sep: I translated the 1931 seminal paper Graphs and Matrices by Dénes Kőnig from Hungarian to English.
- Aug: I became a steering committee member of the Linked Data Benchmark Council.
- Aug: Our paper Towards Testing ACID Compliance in the LDBC Social Network Benchmark was accepted at TPCTC 2020.
- Aug: I started working at the Database Architectures group of CWI under the supervision of Peter Boncz.
- Jul: LDBC conducted its first successful audit for a new database management system (previous audits targeted systems by vendors within the original LDBC European Union project). I supported the implementation process and supervised the audit.
- Jun: I presented our paper Supporting Dynamic Graphs and Temporal Entity Deletions in the LDBC Social Network Benchmark’s Data Generator at the GRADES-NDA 2021 workshop of SIGMOD.
- Mar: Our paper An incremental GraphBLAS solution for the 2018 TTC Social Media case study was accepted at the GrAPL 2020 workshop of IPDPS.
- Feb: I visited CWI in Amsterdam with Benjamin Steer and Jack Waudby. We worked with Peter Boncz on the LDBC Social Network Benchmark.
- Feb: I gave a talk at FOSDEM’s High-Performance Computing developer room about GraphBLAS.
- Jan: I was a program committee member of the EDBT 2020 conference.
- Jan: Dominik Tomaszuk visited me at BME. We worked on surveying existing graph data models and schema languages.
2019
- Dec: I participated in the Dagstuhl seminar on Big Graph Processing Systems.

Notes

Using the Aztec Code on an iPhone to enter the gates at Dutch train stations

TL;DR: To prevent your iPhone from switching to Apple Pay at train station gates when scanning your ticket’s code, use Guided Access mode. This can be turned on in the accessibility settings and activated by triple-clicking the side button of your phone. Recently, when travelling to FOSDEM from Amsterdam to Brussels via railway, I ran into the following problem. The Aztec Code for my train was saved on my iPhone. When I pulled out the phone to scan it at the entrance gates of Amsterdam Centraal, my phone kept opening the wallet for Apple Pay, making the Aztec Code no longer visible....

Publications at VLDB 2023

I co-authored the following papers, to be presented at VLDB 2023 and its satellite events. These papers capture the work on the new and updated benchmarks of the Linked Data Benchmark Council (LDBC), as well as LDBC’s organizational restucturing. Additionally, an initial DuckDB-based prototype of the SQL/PGQ language extension was accepted at the demo track. VLDB 2023 main track – G. Szárnyas et al.: The LDBC Social Network Benchmark: Business Intelligence workload TPCTC 2023 – G....

E-bike rental options in Amsterdam

Recently, I had a few family members visiting me in Amsterdam and I wanted to show them around the Dutch countryside. To give us a larger radius – and put towns like Marken, Zaandam, and Naarden comfortably within reach – I decided to rent e-bikes. In this post, I document my findings about the available options and the bikes I rented. Note that my lessons learnt are valid as of summer 2022 and are likely to change in the future....

Self-joins

Introduction A self-join is a relational join operation which joins a table to itself, typically based on equivalence predicates. For example, given the table \(\mathsf{knows}(\mathsf{person1\_id}, \mathsf{person2\_id})\), representing a social graph where people know each other, and aliases of this table \(\mathsf{knows1}, \mathsf{knows2}, \ldots\), take the following query: SELECT knows1.person1_id AS pA, knows1.person2_id AS pB, knows2.person2_id AS pC FROM knows knows1 JOIN knows knows2 ON knows2.person1_id = knows1.person2_id; The query specifies a self-join on the \(\mathsf{knows}\) table, \(\mathsf{knows1} \bowtie \mathsf{knows2}\), and produces two-hop paths in the social graph....

Graph query languages

The maze of graph query languages can be difficult as there are a number of similarly named languages with subtle differences. This document contains a very brief description of popular graph query languages. Languages Property graphs A number of languages exist for defining graph queries over property graphs. Here’s a list of 10 popular languages: Cypher: The declarative graph query language used by the Neo4j graph database. Cypher popularized the ASCII art syntax for defining graph patterns....

News#

Notes#

News

Notes