Post by Grab

Understanding the behaviour of millions of users requires massive data. Protecting that data requires robust architecture. To scale privacy alongside innovation, our engineers built a dedicated PII Masking Service. 🛡️ It acts as an automated firewall between raw production streams and our development teams. Here is how we engineered it: - The Gatekeeper: A custom CI pipeline scans every new Protobuf schema. If a sensitive field lacks a PII tag, the build halts automatically. - The Transformer: An in-house Flink application intercepts production Kafka topics. - The Mask: Using consistent HMAC hashing, the system converts personal details into anonymised tokens before they ever reach the staging environment. This allows our Data Scientists to train models on realistic behaviour patterns—without ever accessing a single real name or phone number. Privacy isn't just a policy here. It’s part of the codebase.