GenAI Cracow #30: Neuro-symbolic AI
Neurosymoblic Ai has been a growing trend in 2026. While "ai influencers" brag on how many tokens they burned in claude code, building another CRM no-one will use, we
Agenda
Opening Ceremony
"More solvers, less tokens" by Konrad B
"Student-Course Assignment with MiniZinc" by Konrad M
Networking
Abstract
More solvers, less tokens
Two nested loops, a clever trick, a chat window three ways to be wrong. The same problem at sudoku, invoice reconciliation, assembly line sequencing, wearing different spreadsheets. What if we write down the rules instead of the prompt, then watch a solver work on it?
Student-Course Assignment with MiniZinc
Assigning about 100 students per cohort to 10–20 elective courses used to be done by hand at my faculty's dean's office; this year a MiniZinc model running on CP-SAT did it for six cohorts. A web form collects each student's course preferences and friendships, then the model minimises objectives one stage at a time — unmet preferences first, then grade averages, broken friendships, and lecturers' teaching hours — locking each result before the next. The same constraint modules also validate a hand-edited assignment and repair an invalid one with the fewest possible changes; a full solve takes under 10 seconds.
Speakers
Konrad Bujak
Farmer, AI Engineer & Venture Builder
LinkedIn
Konrad Małek
Computer Science and Intelligent Systems student and Junior Software Engineer at Codibly. Interested in Artificial Intelligence, optimization and computer vision. After hours: sailing and skiing.
LinkedIn
Partners
hub:raum
CEE AI Hub