Agentic AI
Human approval is a feature, not a fallback
Bounded agent runs, risk classification, and approval gates for autonomous software delivery.
Read insight →AI engineering insights
Notes from production systems.
Practical perspectives on AI, cloud, security, and regulated software—grounded in systems I have built and led.
LLM operations
Controlling LLM cost without slowing delivery
Caching, quality evaluation, fallback models, and tenant-aware spend controls.
Read insight →AI governance
Securing enterprise AI coding workflows
Managed policies, permission boundaries, and platform-aware sandboxing.
Read insight →Healthcare AI
Building AI platforms around privacy boundaries
Tenant isolation, auditability, and access controls in regulated platforms.
Read insight →Identity security
Enterprise biometric authentication
Deployment choice, adaptive matching, liveness, and privacy controls.
Read insight →TaxTech
High-throughput systems with accuracy at scale
Rule engines, vectorized pipelines, encryption, and compliant delivery.
Read insight →AWS
High-traffic content platforms on AWS
Separate delivery and publishing to improve scale and operational speed.
Read insight →Healthcare technology
Reliable healthcare booking systems
Real-time availability, concurrency control, encryption, and audit trails.
Read insight →