Principal Edge AI Infrastructure & Agentic Platform Engineer

About the Role

We are seeking a highly technical, hands‑on Principal Edge AI Infrastructure Engineer to bootstrap, secure, and operate a bare‑metal edge AI platform from the ground up.

In this role, you won't be inheriting a polished, out‑of‑the‑box cloud environment. Instead, you will be the foundational engineer responsible for getting your hands dirty: manually bundling dozens of open‑source components, testing compatibility, and managing the upgrade cycles until you build the automation.

You will own a massive operational surface area spanning physical hardware, VM management, Kubernetes, high‑volume data ingestion, AI inference services, and the agentic workflow layer that allows AI to interact safely with our systems.

Bootstrap and operate a highly secure, private edge AI platform where infrastructure, data, AI inference, and autonomous agent workflows can be bundled, deployed, continuously maintained, and governed by strict zero‑trust policies from day one.

Servers, GPUs, firmware, Linux, storage, networking

VM lifecycle, hypervisors, templates, snapshots, HA, isolation

OSS component bundling, version compatibility, manual/automated patching, upgrades

Monitoring, incident response, backup/recovery, capacity planning, runbooks

• Bootstrap the edge AI stack from scratch, manually bundling and integrating 40+ open‑source infrastructure, data, and machine learning components.
• Own the patch management and upgrade cycles for all layers (OS, hypervisor, Kubernetes, data stores, and AI runtimes), diligently testing compatibility before rolling out to production.
• Progressively build the automation and Infrastructure‑as‑Code (IaC) required to transition from manual integrations to repeatable, push‑button deployments.

The candidate will own the virtualized infrastructure layer used to support edge AI operations, security isolation, management services, agent runtimes, and non‑containerized workloads.

• Design and operate the virtualization layer across edge and bare‑metal sites.
• Manage hypervisors (e.g., VMware vSphere/ESXi, Proxmox, KVM/libvirt, Open Stack, Harvester, Kube Virt).
• Create and maintain golden VM images, templates, cloud‑init configs, and hardened OS baselines.
• Allocate CPU, memory, storage, GPU, and network resources across VMs and Kubernetes workloads, supporting hardware acceleration (GPU passthrough, vGPU, SR‑IOV) where applicable.
• Design VM networking with VLANs, virtual switches, firewall zones, and management‑plane isolation.
• Define intelligent workload placement rules across bare metal, VMs, Kubernetes, and isolated agent sandboxes.
• Maintain runbooks for failure recovery, hypervisor patching, node evacuation, and disaster recovery.

The candidate will build and operate the workflow layer that allows AI systems, operators, and applications to safely trigger tools, workflows, approvals, data lookups, optimization proposals, and infrastructure actions.

• Design and operate AI tool registries, agent gateways, and Model Context Protocol (MCP) servers.
• Build agentic workflows for read, propose, approve, apply, monitor, and rollback patterns.
• Ensure AI agents do not directly mutate infrastructure or mission‑critical business data without deterministic policy checks (e.g., OPA/OpenFGA) and approval gates.
• Implement human‑in‑the‑loop approval workflows for sensitive AI actions.
• Operate isolated agent sandboxes for the safe execution of tools, scripts, diagnostics, and remediation jobs.
• Build workflow orchestration using tools such as Airflow, Argo Workflows, or Lang Graph.
• Integrate agent workflows with infrastructure, data, and business operations (e.g., health checks, incident triage, data quality diagnostics, compliance queues).
• Maintain absolute auditability: log every agent action with the actor, tool, payload hash, approval state, and execution result.
• Prevent raw tokens, PII, restricted payloads, and sensitive credentials from entering LLM…