Data Breakthroughs - Episode 8: The Office Kitchen Paradox

Real-world data problem solving in action! Irena Bojarovska and host Lior Barak tackle a community-submitted challenge for the first time during the recording.

Problem Category: Machine Learning & AI Implementation Runtime: 50 minutes

The Challenge: A hackathon team built a smart kitchen demand forecasting model with 91% accuracy, but the company is still throwing away 20-25% of fresh products weekly while running out of popular items.

The Solution: The breakthrough isn't about fixing the model, it's about fixing the data. The model is missing critical inputs (office attendance, special events) and is operating blindly due to data quality problems. The real solution combines better data, human-AI collaboration, and proper A/B testing.

Key Takeaways: • Model accuracy ≠ real-world performance (91% test accuracy doesn't guarantee waste reduction) • Data quality and contextual information are your foundation (garbage in, garbage out) • Humans should augment the model, not be replaced by it (hybrid approach wins)

Guest: Irena Bojarovska, Data Scientist at Zalando SEConnect: https://www.linkedin.com/in/irenabojarovska/

Get Involved: Submit your data problem: https://data-breakthroughs-podcast.cookingdata.blog/submit-problem Become a guest: https://data-breakthroughs-podcast.cookingdata.blog/become-guest Join the conversation: #DataBreakthrough

Full show notes & visual diagrams: https://wabi-sabi-data-newsletter.com [or your actual newsletter link]

Figma Board: https://www.figma.com/board/jfC4ipNvd8zSPIyZreEten/Irena-Bojarovska?node-id=1-14&t=Q46O2Ae9yuRHZRwy-1

Disclaimer: This podcast is for inspiration and educational purposes. Solutions discussed are general approaches; adapt them to your specific context and constraints.

Music: "Calisson" courtesy of Riverside

Get full access to Cooking Data guided by Lior at cookingdata.substack.com/subscribe