This is a specialized seat focused on the silicon, hardware, cloud, inference, and security platforms that Deepgram's voice AI runs on - the partners whose chips, servers, accelerators (CPU/GPU/NPU), cloud and inference platforms, on-device and edge runtimes, and confidential-computing or model-security layers determine where and how our models can be deployed. Working understanding of deployment and infrastructure: containers and orchestration (Docker, Kubernetes/Helm), inference on GPUs/accelerators, and the trade-offs across self-hosted, on-prem, air-gapped, dedicated, and edge/on-device deployments - including basic latency, throughput, and benchmarking concepts.