← BackResume ↗

Case Study

AI Surveillance: Anomalous Trade Detection

Improved surveillance accuracy and confidence scoring to flag high-risk trades early.

AI/MLMarket SurveillancePrecision/Recall

35%

Accuracy uplift

-22%

Alert review time

100K+ users

Adoption

Problem

Surveillance analysts were overrun by noisy alerts, making it harder to isolate real risks fast.

Approach

Re-tuned detection models, introduced explainability layers, and partnered with compliance to validate guardrails.

Outcome

Accuracy improved 35% with higher analyst trust and faster triage cycles.

What I learned

Guardrails plus transparent scoring unlock adoption faster than raw model gains alone.

What I'd do next

Expand to cross-asset correlations and real-time feedback loops for continuous learning.

How I work