RentSignal APIs

Rental platforms face a fraud pattern regular e-commerce doesn't — items can be requested and never returned, or accounts can be used to cycle through rentals with no intent to pay. RentSignal APIs power the intelligence layer behind an e-commerce rental platform to address that: recommendations and fraud-risk scoring for rental requests, delivered as a clean API suite the main platform consumes, tuned to flag risk without blocking legitimate customers.

What I Built

• Recommendation API suggesting items based on browsing and rental history
• Fraud-risk scoring flagging suspicious rental requests before approval
• Dynamic pricing signals based on observed demand patterns
• Clean REST integration consumed directly by the main platform

Tech Stack

• Backend: Python, Flask
• ML: scikit-learn for risk scoring
• Database: PostgreSQL
• API: REST

Key Decisions

• Shipped the intelligence layer as a clean REST API suite the main platform consumes, rather than a rewrite of the platform itself, so recommendations and fraud scoring could ship without disrupting the existing rental system
• Tuned the fraud-risk model to flag suspicious requests without blocking legitimate customers, since rental fraud patterns (no-return, pay-to-cycle accounts) needed a scoring approach rather than a hard rule that risked false positives

Outcome / Impact

Reduced fraudulent rental requests while improving conversion through better-targeted recommendations — intelligence layered cleanly on top of the existing platform rather than a rewrite.