Method
BiLSTM
Bidirectional LSTM reading sequences both ways.
7 cases from 3 companies apply BiLSTM. 2 were checked in an online A/B test; 1 report results in production.
Typical tasks: Retrieve and rank from millions of items, Reduce ML cost, Decide who gets an offer, Predict churn and lifetime value. Companies: Alfa-Bank, Target, Zomato. Often combined with: A/B test, Regular expressions, fastText, Focal loss, Named entity recognition and 25 more.
| TargetRun: AI-Powered Trip Prediction for Delivering Timely Offers | retail | 2026-06 | |
| Applying NLP Models to Automate Call Center Work | banking | 2024-05 | |
| Development History of Voice Bot: Legacy Voice Bot of Alfa-Bank and Transition to Modern Methods | banking | 2023-05 | |
| NLP at Alfa-Bank, Part 9: Structuring OFD Receipts | banking | 2023-05 | |
| Predicting your order's Food Preparation Time | food-delivery | 2022-07 | |
| Neural Network Models on Card Transactions | banking | 2022-06 | |
| Churn and Propensity Tasks on Neural Network Transaction Models | banking | 2022-06 |