How to build a secure RAG pipeline on AWS
Security controls for retrieval, embeddings, secrets, model access, logging, and network boundaries.
Writing
Articles for operators building AI in the real world.
These pieces are designed to teach something useful, show how I think, and create a public record of practical work in AI security, cloud architecture, and compliance.
AI Security
Cloud Architecture
Compliance
Why most AI systems fail compliance (and how to fix it)
The common reasons AI products break under ISO 27001 and SOC 2 scrutiny, and the design moves that help.
Local LLMs vs cloud LLMs for regulated environments
How to think about data residency, threat surface, model operations, and where each option fits.
Designing ISO 27001-ready AI infrastructure
What should be true about access, asset boundaries, recovery, evidence, and change control.
The biggest security risks in modern startup architecture
The failure modes I see most often in fast-moving engineering teams and how to address them early.