Building a secure data platform: why good design and security go hand in hand
summary
Claims that Australian breaches cost about 3.35 million AUD and rise 9.8% year on year drive concrete steps: classify at onboarding with simple tiers or PII/non PII, record data contracts (YAML AVRO or JSON) with owner and rationale, enforce landing and encrypted zones via VPC subnets private endpoints and endpoint policies, use managed identities and SCIM for deterministic roles, apply masking tokenization or vault tokenization by frequency and geography, redact logs, track lineage via the contract and use the Practical Data Testing Grid to avoid real production test data and automate controls.
AI-written summary · about
chapters (22)
- 0:00Introduction
- 1:50Poll
- 4:02Onboarding
- 5:03Motivation
- 6:44Classification
- 8:20Capture
- 8:34Data contracts
- 10:30Contract example
- 11:51Basic cloud security
- 12:23Private subnets
- 14:58Data zones
- 16:02Encryption
- 16:29Tokenization
- 17:32Injection and Storage
- 18:46Parque
- 19:05Processing
- 22:26Data
- 25:29Data Lineage
- 26:26Summary
- 27:02Data consumption
- 30:13Key takeaways
- 31:29Guiding principles
related talks
01h 59mDeep Dive Workshop – Data Security and Data Protection
40mSecure application design with high data privacy requirements
16mSecurity, Fraud, and Compliance
45mKeeping Your AI/ML Data Secure
43mSecurity By Design
37mAdd security to architecture one step at a time
01h 59mDeep Dive Workshop – Data Security and Data Protection
28mBreaches Are Everywhere; What's a Good Security Leader to Do