PhilippinesFinance and insurance2025 – ongoingMaya
Machine learning for underwriting, fraud detection and personalization
Maya reported a 45% rise in revenue per user from AI recommendations and up to 98% lower fraud losses
Problem
Maya, a Philippine fintech, serves customers with varied financial histories. It needed to judge credit risk, stop fraud and target offers across its banking app.
Approach
Maya built machine learning models on the Databricks platform, with MLflow managing the model lifecycle. The models set credit limits from customer behavior, transactions and device use. Fraud detection combined transaction sequence embeddings, network analysis and predictive models. AI-driven personalization built customer profiles for campaigns, and optimization was automated for hundreds of marketing campaigns a month.
Results
Takeaway
Maya applied the same behavioral data to lending, fraud and marketing. The page gives the fraud result only as an upper bound.
Categories
- Industry
- Finance and insurance
- Business function
- Finance, risk and legalMarketing and sales
- Innovation type
- Product and service innovation
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Databricks describes how Maya uses machine learning for underwriting, fraud detection and personalization. It reports $2.1 billion in loans dispersed, fraud losses cut by up to 98% and a 45% rise in revenue per user from AI recommendations. The page gives no start year or staff count.
Full details are in the original sources.
