Our Student Reaches the Podium at Since AI Hackathon 2026
Tomasz Życki, a student at the Faculty of Mechatronics of the Warsaw University of Technology, won third place at the Since AI Hackathon 2026, held in early November in Turku, Finland. The annual event is organised by Since AI, a non-profit organisation founded by students of the University of Turku. The hackathon lasted 72 hours and brought together more than 1,000 participants from around the world, including software developers, AI specialists and technology professionals.
Participants worked on real-world challenges submitted by 15 companies, with each company presenting its own competition category. Tomasz secured third place in the challenge set by Europress, a Finnish manufacturer of waste management and recycling equipment.
“The challenge was called SMART Material Value Engine,” explains Tomasz Życki from the Faculty of Mechatronics. “Europress offers the SMART system, which connects equipment to the cloud and continuously reports data such as fill levels. The task was not to build yet another application, but to propose a new business model: how to use the data from SMART devices to move from selling machines to selling services.”
Tomasz developed a solution consisting of three key components:
- 1. Weighing waste without scales on every machine.
Installing weighing systems on every piece of equipment is costly. Tomasz proposed using a limited number of actual weight measurements to train a model, which would then enable the SMART system to estimate the weight of each collection based on fill-level data. In simulation tests, the estimation error was reduced from 18.7% to 8.7%.
- 2. BinSight, or a monthly bill for customers
The BinSight tool combines data from SMART devices with waste collection invoices and identifies potential discrepancies, such as charges for collecting partially filled containers. Each result is accompanied by supporting evidence and a confidence level. The calculations are performed using conventional software, while artificial intelligence is used solely to explain the results in accessible language, but the final decision is always made by a human.
- 3. Integration with the customer’s AI tools
The solution also includes an Ask AI feature, allowing users to query their own data, as well as an MCP server with pre-built workflows. This enables organisations to connect SMART data directly to AI assistants they already use within their business processes.
“I participated on my own,” says Tomasz Życki. “During the hackathon, I developed the calculation engine, backend, frontend, business case and demo video. I really enjoyed both the location and the atmosphere of the event. The mentors and people I met there were extremely supportive. Working on a real business challenge, with access to company data and subject-matter experts, is a completely different experience from a typical project.”
The proposed solution could deliver tangible benefits to a wide range of stakeholders. Europress, obtains an idea for a new revenue stream based on data-driven services rather than solely on equipment sales. For customers, including retail chains, manufacturing facilities and logistics centres, it provides a clearer understanding of the amount of waste they generate, makes it easier to identify invoicing errors and helps reduce operational costs. More accurate waste-mass data can also support recycling reporting and help reduce unnecessary waste collection journeys, contributing to lower CO₂ emissions.




