Wodan AI

The problem

ML platforms leak in ways traditional AppSec does not catch — training data exfiltration, model inversion, prompt injection, adversarial examples. Security teams do not have tooling built for ML threat models; ML teams do not have security engineering discipline. Both sides need better instruments.

What we built

AR Data contributed engineering on Wodan's ML security platform — telemetry into training and inference, adversarial detection, model access controls, and data lineage. Bridges the gap between ML operations and security operations.

Results

• ML-specific threat detection
• Training + inference telemetry
• Model access controls
• Data lineage tracking

Tech stack