Domagoj Bradač

Domagoj Bradač is a Croatian mathematician working in extremal and probabilistic combinatorics, currently a postdoctoral researcher in the Extremal Combinatorics group of Oliver Janzer at EPFL. He completed his PhD at ETH Zürich under Benny Sudakov. In May 2026 he posted a breakthrough on off-diagonal Ramsey numbers, lowering the bound r(s,k) to match the 90-year-old Erdős–Szekeres upper bound up to polylogarithmic factors and sharpening it to the conjectured exponent k^{s-1+o(1)} — a result described in the combinatorics community as settling a problem open for ~70 years. Before his research career he was an elite competitive programmer, winning IOI silver medals for Croatia in 2015 and 2016 and competing at the ICPC World Finals in 2017 and 2019.

Signal

Primarily an academic communicator via papers, talks and seminar announcements rather than social media; his public footprint is research-led (arXiv, Scholar, conference pages, university pages) with high-school-era Croatian media coverage of his olympiad medals.

Notes14

• Posted 'Off-diagonal Ramsey numbers' (arXiv 2605.28793) on 27 May 2026, proving r(s,k) >= Omega(k^{s-1}/log^{2s-4}k) for fixed s>=3 — the first lower bound matching Erdos and Szekeres' 90-year-old upper bound up to polylogs; for s>=5 it breaks Spencer's 1977 k^{(s+1)/2+o(1)} bound, which suggests this is a landmark result rather than an incremental one.May 27, 2026
• Within three weeks (updated 16-17 June 2026) he sharpened the Ramsey exponent from s-2 to the conjectured s-1, i.e. r(s,k)=k^{s-1+o(1)}; a blogger calls it 'after 70 years we finally have the correct exponent, which was already bravely conjectured by Erdos in 1947' — unusually fast consolidation of a headline result by a single author.Jun 17, 2026
• Now a postdoc in Oliver Janzer's Extremal Combinatorics group at EPFL (Lausanne), listed as 'Scientist' and also a 'Lecturer' teaching the MATH-360 Graph Theory course — research and teaching split.Sep 1, 2026
• PhD at ETH Zurich's Mathematics Department advised by Benny Sudakov, a leading extremal-combinatorics group; his webpage labels him a 'fourth year PhD student', which dates his PhD cohort to ~2021-2022.May 1, 2026
• Corroborated academic email [contact omitted] plus a public office phone [private detail omitted] on the EPFL directory page.
• Timeline synthesis: IOI silver 2015-16 -> undergraduate contest era (COCI problem authorship 2016-17) -> ICPC World Finals 2017/2019 -> ETH PhD (~2021) -> EPFL postdoc -> 2026 Ramsey breakthrough. Each stage hands off to the next within ~2-3 years, i.e. a continuous, unusually steep research-competition trajectory with no gap years.
• The 2026 Ramsey result is single-authored while nearly all his prior work is co-authored with Sudakov/Janzer and the ETH-EPFL combinatorics cluster; a lone authorship on the biggest result suggests he drove the key idea himself rather than extending an advisor's framework.
• The breakthrough was noted on Gil Kalai's blog 'Combinatorics and more' (June 2026), placing him in the small circle of researchers whose single-author results are tracked by senior theorists.Jun 6, 2026
• Authored IOI-style contest problems including 'Wells' (CEOI 2021, Day 2) and 'Burza', 'Kralj' (COCI 2016/2017), 'Izlet', 'Tenis' (Croatian Olympiad in Informatics 2019) — a give-back to the Croatian/CEI competitive-programming pipeline he came up through.
• Google Scholar metrics as of Sept 2026: 208 citations, h-index 9, i10-index 8, with 198 citations since 2021 — nearly all impact concentrated in the last five years, consistent with an early-career researcher on a steep curve.Sep 11, 2026
• Two 2026 papers with Oliver Janzer ('Hamiltonicity of regular sublinear expanders', May 2026) indicate his new EPFL group is already generating joint output — the Janzer collaboration postdates and extends the ETH-era co-authorships.May 14, 2026
• Early work (2019-2020) was in theoretical CS/online algorithms ('Robust Algorithms for the Secretary Problem', ITCS 2020; '(Near) Optimal Adaptivity Gaps for Stochastic Multi-Value Probing', RANDOM 2019, with Sahil Singla and Goran Zuzic) before pivoting fully to combinatorics — reflective of a TCS-to-pure-math trajectory.Jan 1, 2020