Principal Infrastructure Engineer

About

Edison Scientific builds and commercializes AI agents for science. Scientific discovery moves too slowly, and autonomous AI agents are how we intend to fix that. We’re assembling a team of top researchers and engineers across AI and biology to build an AI scientist.

As a Principal Infrastructure Engineer, you’ll play a key role in designing, scaling, and operating the core platform infrastructure that powers autonomous scientific discovery. Your primary focus will be the orchestration for our agents at scale — building and managing clusters that orchestrate thousands of persistent, stateful workloads, developing custom resource definitions (CRDs) and operators, and ensuring the reliability and efficiency of our compute layer at scale.

Our mission is to build an AI scientist, and you’ll own the infrastructure foundation it runs on. AI agents performing long‑running scientific research demand resilient scheduling, lifecycle management, and resource orchestration far beyond typical cloud‑native workloads. This role will influence platform architecture, establish infrastructure best practices, and partner closely with backend engineers, ML engineers, and researchers to deliver a production‑grade environment that lets science move faster.

At Edison Scientific, engineering at the senior level is about technical ownership and leverage—understanding how complex systems interact, making sound architectural tradeoffs, and building foundations that allow teams and science to move faster.

This role is on‑site at our San Francisco office in the Dogpatch neighborhood. Our office is a converted warehouse with high ceilings, open space, and a team that genuinely believes in what they’re building.

This position is part of the Platform team.

• Architect, implement, and operate Kubernetes clusters that support thousands of concurrent, persistent resources (agents, jobs, services) with high availability and efficient resource utilization.
• Design and develop custom resource definitions (CRDs) and Kubernetes operators to model and manage domain‑specific workloads such as AI agent life cycles, research pipelines, and long‑running compute tasks.
• Drive the strategy for cluster scaling, node pool management, autoscaling policies, and resource quota frameworks to handle rapid workload growth.
• Build and maintain infrastructure‑as‑code (Terraform, Pulumi, or similar) for reproducible, version‑controlled environment management.
• Design and implement robust scheduling, placement, and affinity strategies to optimize cost, performance, and fault tolerance for heterogeneous workloads (CPU, GPU, memory‑intensive).
• Establish and uphold best practices around observability, monitoring, alerting, and incident response for infrastructure systems (Prometheus, Grafana, Datadog, or similar).
• Own storage and networking strategy within Kubernetes — including persistent volume management, CSI drivers, service mesh, network policies, and ingress architecture.
• Troubleshoot complex, cross‑system infrastructure issues and guide others through effective debugging and remediation in distributed environments.
• Collaborate closely with backend, ML, and research teams to understand workload requirements and translate them into reliable infrastructure patterns.

• 10+ years of professional infrastructure or platform engineering experience, with deep hands‑on Kubernetes expertise in production environments.
• Experience designing and implementing custom resource definitions (CRDs) and Kubernetes operators (using frameworks such as Kubebuilder, Operator SDK, or controller‑runtime).
• Track record of operating and scaling Kubernetes clusters supporting thousands of persistent or long‑lived resources (stateful workloads, persistent pods, long‑running jobs).
• Deep understanding of Kubernetes internals—API server, etcd, scheduler, controller manager, kubelet—and how they behave at scale.
• Expertise with cloud infrastructure (AWS EKS, GCP GKE, or Azure AKS) and associated networking, storage, and IAM primitives.
• Proficiency in at least one systems or backend language for operator development and infrastructure tooling.
• Hands‑on experience with…