AutumnTHE OLYMPIAD COLLECTION
Menu
RESEARCHED WITH AUTUMN
MT

Martin Töpfer

AI engineer/researcher @Stealth | formerly at Everpure (Pure Storage) & Qminers
Stealth AI Safety Startup · Praha - metropolitní oblast · engineer
IMO 2012 Honourable Mention (Czech Republic)MSc Computer Science, Charles University (2019); Thesis On Intersection Representations Of GraphsQminers (quantitative Trading, Prague) Python Developer, 2019-2020Pure Storage Member Of Technical Staff, 2021-2026Member Of Technical Staff At A Stealth AI-safety Startup, 2026-present

Martin Topfer is a Prague-based software engineer and researcher. He studied Computer Science at Charles University in Prague (BSc 2012-2015, MSc 2015-2019, thesis 'Intersection representations of graphs') and represented the Czech Republic at the 2012 International Mathematical Olympiad (Honourable Mention). After two DIMACS research stints at Rutgers University and an unfinished PhD at the Institute of Science and Technology Austria (population protocols, 2017-2019), he worked as a Python developer at the Prague quantitative trading firm Qminers (2019-2020), then spent about five years as a Member of Technical Staff at Pure Storage (2021-2026) on testing infrastructure, and since 2026 works on AI safety at a stealth startup. His headline self-describes as 'AI engineer/researcher'.

I like coding when I do science and I like science when I code.

Details
LocationPraha - metropolitní oblast
UniversityCharles University (Univerzita Karlova)
Pythonalgorithmsgraph theorydiscrete mathematicssoftware testing infrastructuremachine learning
Notes
  • Quant employment verification: CONFIRMED. Martin Topfer worked as a Python Developer at Qminers, s.r.o. (Prague) from June 2019 to November 2020 (LinkedIn index: 2019-2020). Qminers is verified as an actual quantitative trading firm, not a name match: its own site describes it as 'a quantitative trading firm operating on global markets, with strategies built end to end in-house', ~50 people, citing a 10 Sharpe ratio and a partnership with a major US trading house. His LinkedIn describes the role as backend Python development in 'a futures exchange company' - internal tooling, performance work and exchange-data analysis - i.e. a software-engineering role inside a quant firm, not a quant researcher/trader. Identity is supported by his Charles University master's thesis, the DIMACS REU page and the same-name IMO 2012 record, all consistent with one Prague-based CS graduate.Jun 2019
  • Currently a Member of Technical Staff at a stealth AI-safety startup in Prague since 2026 (title/company 'Stealth AI Safety Startup', no public domain); LinkedIn last active 2026-09-04.2026
  • Was a Member of Technical Staff at Pure Storage (Everpure), 2021-2026 (~5 years), as a Python software engineer on testing infrastructure (extensible design, efficiency, robustness, scaling) in Prague; his own headline reads 'AI engineer/researcher @Stealth | formerly at Everpure (Pure Storage) & Qminers'.2021
  • His career reads as research-first then industry: Charles University CS (BSc 2012-2015, MSc 2015-2019) with two Rutgers DIMACS REU stints and papers, an unfinished ISTA PhD (2017-2019), then a pure-software role at the quant trading firm Qminers (2019-2020), ~5 years at Pure Storage (2021-2026) and now a stealth AI-safety startup (2026-) - the only quant-trading employer on the timeline is Qminers, and his role there was engineering rather than trading or research.2026
  • Earned a Master's degree in Computer Science at Charles University (Univerzita Karlova, Prague), 2015-2019, preceded by a Bachelor's in CS there, 2012-2015; thesis 'Intersection representations of graphs' supervised by Vit Jelinek.May 2019
  • Reached IMO 2012 for the Czech Republic with an Honourable Mention, rank 350 and 10 points (P1 7, P4 3); a same-name olympiad record consistent with the Charles University / Prague CS profile.Jul 2012
  • Began an (unfinished) PhD at the Institute of Science and Technology Austria (ISTA), 2017-2019, in Dan Alistarh's group, working on robust algorithms in population protocols - a research detour between his master's and industry.2017
  • Co-authored several papers: 'On H-topological intersection graphs' (arXiv 1608.02389, with Jelinek et al., later in Algorithmica), 'On grounded L-graphs and their relatives' (1808.04148), 'Squarability of rectangle arrangements' (1611.07073), 'Robust Comparison in Population Protocols' (2003.06485) and the Rutgers-REU work 'On the Simultaneous Minimum Spanning Trees Problem' (1712.00253).2016
  • LinkedIn headline self-describes as 'AI engineer/researcher', and summary 'I like coding when I do science and I like science when I code' - consistent across his research (population protocols) and current stealth AI-safety work; the quant-firm stint at Qminers is the software-engineering phase between them.2026
  • A GitHub account 'mtoepfer' exists under the name Martin Toepfer but belongs to a different person (a German NLP/computer-science researcher, co-author on subject-indexing papers); no GitHub account was linked from or attributable to this subject, so none is recorded.2026
  • Participated twice in the DIMACS Research Experience for Undergraduates at Rutgers University (2015, mentor James Abello; and 2016), a program to which Charles University sends top CS students - early evidence of the research track that ran through his Prague studies.Jun 2015
  • Was an undergraduate Teaching Assistant at Charles University for Discrete Mathematics exercises (90 min/week), 2016-2017, while an undergraduate - a teaching thread alongside the research.2016

Competition record

Czech Republic · IMO

Related collections

Explore the collection