SINCE 2026
SeriousNode
Founder & CTO
- Graphs for AI Memory
- Academia + Industry
- Network Epidemiology on Transit
Founder & CTO, SeriousNode · Ex-Cognee · UP FAMNIT
Hajdu László, Ph.D.
Koper, Slovenia
Turned graph theory into shipped products — fraud detection, pricing optimization, and now AI memory infrastructure.
László Hajdu builds the graph infrastructure companies reach for when AI agents need persistent memory. As founding engineer at Cognee he owned retrieval, ontology, and a Rust orchestrator; in 2026 he founded SeriousNode to take that further. The commercial track record runs deep: fraud-detection systems at two fintechs, advanced analytics at Egis Pharmaceuticals, and active consulting on GraphRAG and agent infrastructure. A Szeged PhD in applied graph theory sits behind it all — mostly as a licence to ship.
- 48,546-node Bay Area transit contact network
- 21demand/capacity epidemic scenarios modeled
- 46works in the COBISS catalogue
- 79Scopus citations, as of Aug 2026
HAPPENING NOW
STANDING TODAY
SINCE 2026
Aster AI
Member of Technical Staff
HOW THEY THINK4 BELIEFS
Four convictions run under two decades of graphs.
THE IMPACT
One graph toolkit, pulled through fraud, transit, epidemics, and AI memory.
EVERYTHING ELSE · 10 WORKS
THE TIMELINE
One graph toolkit, carried from Szeged transit software to memory for AI agents.
The whole record, newest first – tap any moment and Aster tells the story.
2024–PRESENT
Graphs become memory for AI agents
The convergence. As founding engineer at Cognee he built retrieval, ontology, and a Rust orchestrator for semantic memory, then founded SeriousNode in 2026 in the same GraphRAG territory. He first-authored the Bay Area transit-epidemic study in Nature Scientific Reports and keeps publishing network science — the oldest ideas now serving LLMs.
“Open to consulting on GraphRAG, AI memory, agent infrastructure, and applied ML.”
2021–2024
The graph toolkit goes to work in pharma and fraud
The commercial years, run in parallel with his research posts. At Egis Pharmaceuticals he led advanced-analytics modules and delivered the price-optimization model that drove revenue; as CTO at FRISK he supervised a fraud-detection system and pitched investors; and he became a Researcher at InnoRenew CoE working on fraud analytics, IoT pipelines, and virus/influence/churn modeling.
2016–2021
A PhD on how things spread through networks
The doctorate at Szeged abstracted his transit work into network science — influence maximization in the Independent Cascade model, community detection, and fraud in financial networks. In parallel he became a Senior Data Scientist at Andego and, from 2018, an Assistant Professor at Primorska and researcher at InnoRenew. He defended his dissertation in December 2021.
“we're developing Skynet and Terminators out of wood”
2011–2016
Transit software, and the first taste of optimization
The foundation at the University of Szeged: a BSc in software information technology, an MSc in computer science, and — running alongside — work as a software developer who shipped an iOS transit route-planning app and optimization projects for real transit operators. Public transportation as a network to optimize starts here.
“somewhere between artificial intelligence and logic exams it grew on me, and it no longer bothered me that I was a 'geek'. And so I became a computer scientist.”