At Defide, I work with backend and AI teams on a production RAG product, from document ingestion and hybrid retrieval to citations and streamed answers. I review backend work, guide technical decisions and write Python when a problem needs deeper investigation.
I also designed the application and deployment approach for moving customer-specific AI workloads from separate EC2 capacity to shared EKS GPU workers. The infrastructure team owned the EKS platform and most of the Terraform; I contributed application-side Kubernetes and delivery changes.
At Telar, my open-source platform and practical R&D space, I build products including Agent Telar, Telar Academy and Telar Social. I use them to explore tools, memory, retrieval, observability and cost-conscious AI architecture, and to share what I learn through practical guides and runnable examples.
