Method
Label noise detection
Auditing and cleaning noisy labels (confident learning, blind audits).
10 cases from 8 companies apply Label noise detection. 1 was checked in an online A/B test; 4 report results in production.
Typical tasks: Catch fraud and abuse, Moderate content at scale, Detect anomalies, Recommend to new users and new items and 2 more. Companies: Alfa-Bank, Celsus, LY Corporation, DoorDash, Avito, Gusto and 2 more. Often combined with: Threshold selection, LoRA fine-tuning, LLM-as-a-judge, Gradient boosting, Constrained decoding and 33 more.
Developing a model to detect harmful content from OpenChat titles and descriptions![]() | ![]() | social | 2026-08 |
Fox One: Video Content Personalization Powered by GenAI![]() | ![]() | media-streaming | 2026-06 |
| With Eyes Wide Shut: Detecting Sleeping People and Other Remote Verification Attacks Without Consent | fintech | 2026-05 | |
| LLM-as-a-Judge: Evaluating natural language search | food-delivery | 2026-05 | |
| Moderation Platformization | classifieds | 2025-05 | |
| Tackling the Human Cost of R&D Tax Credit Qualification with AI-Powered Models | saas-b2b | 2024-09 | |
| History of the Unified Classifier Development | banking | 2023-05 | |
| NLP at Alfa-Bank, Part 2: Data Labeling | banking | 2023-05 | |
Data-Centric AI: A Review of Methods![]() | ![]() | healthtech | 2023-05 |
A review of data cleaning and annotation quality control methods![]() | ![]() | healthtech | 2022-06 |


