Tool
ViT
11 cases from 8 companies use ViT in their stack. 3 were checked in an online A/B test; 3 report results in production. Tasks: Catch fraud and abuse, Optimize decisions, Learn user and item embeddings, Moderate content at scale and 3 more. Methods: Embedding similarity matching, Human in the loop, Gradient boosting, Triplet loss, Image classification and 40 more. Companies: Avito, Wildberries, DoorDash, VK, Canva, Airbnb and 2 more.
| How Ad Ranking Works at Avito | classifieds | 2026-05 | |
| A Digital Core DocumenterUS | US | manufacturing | 2026-05 |
| Description Generation One Year Later. What We Managed to Understand Over the Year | classifieds | 2025-05 | |
| Developing Multimodal Content Models for Recommendations | social | 2025-05 | |
High-Speed Multimodal Moderation: LLMs and VLMs in Real Production![]() | ![]() | marketplace | 2025-05 |
| Using LLM to transcribe restaurant menu photos | food-delivery | 2025-03 | |
| Image replacement in Canva designs using reverse image search | general | 2025-01 | |
| Part 1: DoorDash 2024 summer intern projects | food-delivery | 2024-10 | |
| Geo Embeddings: Universal Representations for Locations | banking | 2024-05 | |
How We Fought Spam in Reviews![]() | ![]() | marketplace | 2024-05 |
| Building Airbnb Categories with ML & Human in the Loop | travel | 2023-03 |
