SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Mariana Minerals is a software-first, vertically integrated minerals company supplying critical minerals for modern energy, AI, and defense technologies. The company is reimagining the minerals supply chain through advanced software, automation, and data-driven decision-making.
You will join the infrastructure team that owns the platform underlying MarianaOS—the operating system for how Mariana engineers, constructs, and operates its mines and mineral processing facilities. MarianaOS runs in AWS and on edge clusters at production sites, including a copper facility in Moab, Utah and a lithium facility in East Texas. The platform serves construction dashboards, runs GPU inference on site camera feeds, trains models for lithium extraction circuits, and moves plant floor data into a warehouse.
Your team owns the platform underneath all of that: Kubernetes clusters (EKS and on-premises), deployment pipelines, CI/CD, observability, site/edge infrastructure, identity and secrets management, and cloud cost optimization. You will collaborate with machine learning engineers and data engineers on ML compute platforms and data pipeline infrastructure.
The core philosophy is building for other engineers: when an engineer needs a database, storage bucket, dashboard, or alert, they should get it via a documented self-service path they can run themselves. Your success is measured by how much infrastructure work happens without your direct involvement. The team moves quickly, shipping features and iterating based on feedback from people running plants and job sites.
Key responsibilities include: owning the Kubernetes footprint end-to-end (EKS clusters, on-premises site clusters, upgrades, capacity, autoscaling, ingress, DNS, GitOps delivery); building self-service infrastructure so engineers can provision services, databases, preview environments, and scheduled jobs independently; making observability a default so every deployed service emits metrics, ships logs, traces requests, and carries meaningful alerts; extending the platform to on-premises and edge clusters hosting GPU inference and plant data collection over constrained links; owning cloud costs as an engineering problem with visibility and right-sizing; managing identity, access, and secrets so scoped access is provisioned and revoked automatically; and responding to infrastructure incidents by removing the class of failure that caused them, including writing runbooks.
The tech stack includes TypeScript and React on the front end, Node and Python services, Postgres, and AWS. The platform uses EKS with Karpenter, ArgoCD, Helm, Pulumi, Prometheus, Grafana, Loki, Prefect, Snowflake, and growing amounts of LLM and agent tooling in production.
REQUIREMENTS:
- 8-12+ years building and operating production infrastructure with real depth in Kubernetes (you have run and upgraded clusters yourself and debugged multiple layers below the application)
- Strong software engineering fundamentals; you write code comfortable for product engineer review and reach for well-designed tools/abstractions before runbooks
- Hands-on experience with infrastructure as code (Pulumi, Terraform, or similar) and GitOps-style continuous delivery
- Track record of building platforms other engineers adopted willingly (specifics: what you built, who used it, how you know it helped)
- Practical observability experience with Prometheus, Grafana, and log aggregation stacks; ability to get organizations to instrument services
- Comfort operating across a broad surface with incomplete context
- Fluency in Python or TypeScript and enough AWS familiarity to make sound cost and architecture tradeoffs
- Strong plus: experience supporting ML or data-intensive workloads (GPU scheduling, distributed training, pipeline orchestration)
- Strong plus: experience with on-premise or edge infrastructure, industrial networks, or constrained-connectivity environments
- Helpful but not required: startup experience in a rapidly scaling company
- Beneficial but not required: domain experience in manufacturing, industrial automation, mining, or capital projects
