Company
Affirm
5 cases from Affirm (Fintech), each digested from a public talk or engineering post and linked to the original. 2 were checked in an online A/B test; 1 report results in production.
Tasks: Catch fraud and abuse, Moderate content at scale, Build a data platform, Evaluate models, Learn user and item embeddings. Methods: Gradient boosting, Tabular deep learning, Multi-task learning, Rule-based heuristics, Feature selection, Threshold selection and 10 more. Tools: XGBoost and 1 more.
Tech blog: tech.affirm.com
| Composing Fraud Rules at the Speed of the Attack | fintech | 2026-11 | |
| PredictPay: From Hackathon Idea to Smarter Payments at Scale | fintech | 2026-10 | |
| Precompute First | fintech | 2026-10 | |
| The Model Is Not the Modeling | fintech | 2026-09 | |
| Learning Richer Credit Representations with Transformers at Affirm | fintech | 2026-09 |