russell.ballestrini.net/content/pages/cover-letters/2026-01-10-kobold-metals-senior-infrastructure-engineer-russell-ballestrini.rst

KoBold Metals - Senior Infrastructure Engineer
==============================================

:date: 2026-01-10

.. Applied 2026-01-10

To KoBold Metals:

Precision exploration means fewer holes in the ground, fewer strip-mined forests or mountains. Better data & smarter infrastructure means less environmental destruction & more ethical sourcing of the materials the energy transition needs. That's the impact I want to have.

My infrastructure background maps directly to this role:

**Observability & reliability**: At Remind, I introduced Datadog with EC2 agents for container APM, logs, & custom metrics across 60+ microservices supporting 30M users. Built monitoring & alerting strategies that maximized uptime during COVID-19 lockdowns. I understand what it takes to make systems observable & debuggable for teams who aren't infrastructure specialists, like material scientists focused on discovery rather than debugging.

**MLOps & data infrastructure**: At Coursemojo (2022-2025), I scaled AI-powered educational services, designed CI/CD pipelines for model deployments, & built the infrastructure layer that let data scientists iterate quickly. Managed Elasticsearch clusters for search & analytics workloads. Experience with JupyterHub-style shared compute environments where multiple users need reliable, isolated access to resources.

**Infrastructure-as-code & automation**: CloudFormation, Terraform, Ansible, SaltStack across my career. At Cofense (2020-2022), worked in security-focused environments (SOC2/FedRAMP) where infrastructure needed to be reproducible, auditable, & secure by default. Python is my primary language (15+ years), which aligns well with the scientific computing ecosystem.

**AWS at scale**: ECS, EC2, RDS, DynamoDB, SQS, Lambda. Built infrastructure that handled millions of concurrent users & needed to be both cost-effective & reliable.

**Current project - unsandbox.com**: Building a multi-tenant remote code execution PaaS supporting 42+ programming languages. Similar challenges to JupyterHub at scale: container orchestration, user isolation, resource quotas, & cost optimization through freeze/unfreeze (pause idle containers, wake instantly on demand). Already running Python as persistent services; could spin up on-demand Jupyter notebooks or better alternatives. Container snapshots for reproducible environments & async job management with cancellation. The kind of shared compute infrastructure where scientists submit workloads & expect them to just work.

I also care about what happens after extraction. Watershed management first: retention ponds on contour & swales to sink water into soils, reducing runoff. Then excavation sites restored with native species, food forest crops, & animals, reintegrated with perpetual human management & tending to the people where field operations are conducted.

Russell Ballestrini

https://russell.ballestrini.net | https://github.com/russellballestrini