EdTech Trust & Safety Risk Scoring Platform

The Trust & Safety and Risk teams at an enterprise course creation and creator education platform relied on manual, ad-hoc investigations to flag suspicious accounts for payment abuse, identity fraud, chargebacks, and buy-now-pay-later (BNPL) exploitation. They needed an automated scoring engine coupled with operational dashboards to surface high-risk accounts systematically.

• Co-designed the initial database schema, entity-relationship diagrams, and scoring data models from day one.
• Successfully proposed and locked in an unbounded, additive scoring methodology over a rigid 0–100 scale, preventing score saturation and maintaining mathematical composability.
• Owned client-facing technical discovery and execution: defined integration requirements for payment verification and external fraud-detection APIs, resolved warehouse permissions, and coordinated Web Application Firewall (WAF) rule adjustments.
• Designed and estimated task breakdowns covering all 7 weighted risk dimensions: Payment & Financial Risk, Identity Integrity, Network/Device Risk, Buyer Risk, Content Legitimacy, Reputation History, and Behavioral Sensitivity.
• Implemented background cron workflows, operational risk dashboards with dynamic filtering/search, and administrative controls.

• Additive vs. Bounded Normalization: Defended keeping raw risk points additive in the database and data warehouse rather than clamping scores to a fixed 0–100 range in storage. This ensured historical comparisons remained consistent as new risk signals were introduced, isolating percentage normalization purely to the frontend display.
• Batching vs. Real-time API Quotas: Because the external fraud scoring and identity APIs enforced tight rate limits, scheduled cron jobs were partitioned into micro-batches with retry strategies and allowlisted IP routing to prevent throttling errors.

Shipped the live Risk Dashboard and automated workflows, ingesting and scoring approximately 2,000 accounts with identity and network telemetry at launch. The platform remained in active production operation for months, providing risk analysts with centralized triage.