Customers
Recs Studio is already helping retailers turn their transaction and product data into live, personalised recommendation systems. Here are two production deployments driving real business outcomes.
One of the biggest European retailers
Major Ukrainian retail chain — B2B and B2C e-commerce
Challenge
One of the biggest European retailers wanted to move from rule-based product suggestions to data-driven personalised recommendations across their digital catalogue. The goal was to increase average basket size and improve product discovery without building an internal ML team.
Solution
Recs Studio connected to the retailer's existing data sources, built a TFRS two-tower retrieval model, and deployed a real-time serving endpoint integrated into their e-commerce platform. The full pipeline — ETL, feature engineering, training, and serving — runs in a dedicated Google Cloud project managed by Recs Studio.
Results
- Live personalised recommendations deployed in production
- Retrieval and ranking models trained on historical purchase and catalogue data
- Real-time serving endpoint with sub-100ms latency
- Ongoing retraining schedule keeps models current as the catalogue evolves
Specific business metrics are shared under NDA.
One of the biggest food delivery services in Eastern Europe
Online grocery marketplace in Ukraine
Challenge
One of the biggest food delivery services in Eastern Europe needed a recommendation system that could handle a fast-changing grocery catalogue, seasonal demand spikes, and diverse customer shopping patterns. Accuracy and freshness were critical because grocery customers expect relevant substitutions and complementary products.
Solution
Recs Studio built a multitask recommendation pipeline that combines retrieval and ranking objectives in a single model. Automated ETL refreshes the data, scheduled retraining keeps predictions aligned with seasonal trends, and the serving endpoint scales with demand.
Results
- Multitask retrieval + ranking model in production
- Automated daily data refresh and periodic retraining
- Model serving integrated into the marketplace recommendation slots
- Full observability of endpoint latency, errors, and usage
Specific business metrics are shared under NDA.
Become a Customer
We are onboarding new retailers and B2B marketplaces through the Google Cloud Marketplace. If you are a mid-to-large retailer looking for production-grade recommendations, click Request a Demo or email support@recs.studio.