Tatoi Club
A personalised recommendation system that selects one relevant club service for each member every week and explains the suggestion in Greek or English.
- Role
- AI Engineer, Cloudevo
- Status
- In user testing
- Technology (for technical readers)
- Logistic regression · Beta/Thompson sampling · Gemini · Vertex AI · Cloud Run
~430kpast member interactions analysed
~630club services to choose from
94%selection rate in simulation
505automated checks
How it works
1personalised suggestion per member, each week
- ~430k past interactions
- Understand member preferencesChoose a suitable service
- Balance familiar and new optionsSelect 1 of ~630 services
- Write a short explanationGreek and English; safety checks
- Learn from outcomesUse feedback to improve future suggestions
What I built
- Built a recommendation model using roughly 430,000 past interactions to select a weekly service from around 630 options.
- Added short, personalised explanations in Greek and English, with checks to keep the messages appropriate.
- Tested the approach with 50 simulated members over 10 weeks, recording a 94% selection rate in simulation, alongside comparative testing and 505 automated checks.
- Led the engineering work from early prototype to user acceptance testing.
Could AI solve a problem for your team?
Get in touch