SW Development Engineer 2
About The Position
Dexcom’s Release Engineering team builds cost-efficient AI platforms for product teams and internal developers. The team operates traditional machine-learning and generative-AI models at scale on GCP/GKE, automates model delivery, and manages usage across self-hosted and external providers. It partners with ML Engineering on model quality and training, and AI Engineering on application behavior. This role independently delivers platform components and measurable cost and efficiency improvements. The engineer will build and operate scalable model serving on GKE, including vLLM, and measure and reduce cost per request and token. The role also develops AI gateways with multi-provider integration, cost-aware routing, and reliable fallbacks; automates developer access, budgets, usage limits, and spend alerts; and creates usage and cost reporting with accounting reconciled against provider billing and infrastructure spend. Additional work includes building CI/CD and GitOps automation for model pipelines, registry promotion, validation, deployment, and rollback; managing infrastructure as code; tuning caching, batching, resource allocation, and autoscaling based on load tests and telemetry; and resolving platform incidents with partner teams while improving tests, runbooks, and release reliability. The primary work location is a home office, without an assigned workstation. The employee will coordinate office visits with their manager and must live within commuting distance—typically 75 miles/120 km—of the assigned Dexcom site. Travel is required 10%.
Requirements
- Experience independently delivering production software or cloud-platform components using Python and automated tests.
- Hands-on GCP and Kubernetes experience operating and right-sizing model-serving or compute workloads.
- Experience building CI/CD, GitOps, and infrastructure-as-code automation.
- Strong troubleshooting and collaboration skills.
- Typically, a bachelor’s degree in a technical discipline and 2–5 years of related experience, or a master’s degree and 0–2 years of equivalent industry experience.
- Must live within commuting distance of the assigned Dexcom site, typically 75 miles/120 km.
Nice To Haves
- Experience measuring usage, attributing costs, and improving cloud unit economics.
- Experience with multi-provider gateways such as LiteLLM.
- Experience with vLLM, model registries, or comparable model-delivery systems.
- Experience with infrastructure-as-code tools such as Terraform or Pulumi.
Responsibilities
- Build and operate scalable model serving on GKE, including vLLM, and measure and reduce cost per request and token.
- Implement AI gateways with multi-provider integration, cost-aware routing, and reliable fallbacks.
- Automate internal developer access, team budgets, usage limits, and spend alerts across self-hosted models and external providers.
- Build usage and cost reporting by developer, team, model, and provider, reconciling token accounting with provider billing and infrastructure spend.
- Build CI/CD and GitOps automation for model pipelines, registry promotion, validation, deployment, and rollback; manage infrastructure as code.
- Tune caching, batching, resource allocation, and autoscaling using load tests and telemetry to lower unit cost while meeting model-quality, latency, and reliability targets.
- Resolve platform incidents with partner teams and improve tests, runbooks, and release reliability.
Benefits
- Full and comprehensive benefits program.
- Growth opportunities on a global scale.
- Career development through in-house learning programs and/or qualified tuition reimbursement.
- Opportunity to work at an innovative, industry-leading organization.
Stand Out From the Crowd
Upload your resume and get instant feedback on how well it matches this job.
What This Job Offers
Job Type
Full-time
Career Level
Mid Level
