Member of Technical Staff (Software Engineer, Inference & Training Platform)

Own the infrastructure behind Perplexity’s real-time AI inference and large-scale training workloads. Build and operate a self-serve compute platform that hides GPU provisioning and cluster complexity, schedule workloads to solve GPU scarcity, and ensure both long-running training and low-latency inference run reliably on the same GPU fleet. Drive Kubernetes-based GPU orchestration, fault tolerance, autoscaling, and observability across multiple cloud providers.