I build and run AI infrastructure at Tenzr Health in Vancouver: a HIPAA-compliant RAG platform on Kubernetes, LLM serving on AWS Bedrock, and the CI/CD around it. The numbers below are benchmarked.
Architected and deployed a HIPAA-compliant RAG platform on Kubernetes (FastAPI, Qdrant), with LLM serving on AWS Bedrock and self-hosted models. Own the orchestration, autoscaling, and rollouts at Tenzr Health.
Added PgBouncer connection pooling to the same platform. Connection capacity went up 458%. Wrote the benchmark tool and documentation so the results can be reproduced.
Fixed slow page loads on A2P Energy's marketplace with query optimization and eager loading. Also built a shift-logging system with role-based access and a bilingual interface for industrial users.
Wrote 15+ automation scripts for Azure cloud deployments at SaaSberry, cutting deployment time by 40%. Configured Azure B2C tenants and App Services for government-compliance clients.
Lessons from deploying AI in healthcare
A practical guide to PgBouncer connection pooling
Standardizing code quality without slowing developers down
If you want to talk about infrastructure or performance work, reach me at hi [@] manvirheer [.] com.