Platform Reliability Engineer – AI Infra & GPU Fleet (Remote)

Descript; San Francisco (CA); Hybrid; USD 220,000 - 292,000; Full time

Benefits offered by this job

Catered lunches

Flexible vacation time

401k matching

Remote and hybrid roles

Job summary

Descript seeks an experienced platform engineer to own the foundation of our compute, deployment, reliability, CI/CD, and monorepo health for model training and serving. You will partner with engineers across teams to shape architecture and ensure robust systems.

You will work on on-call processes, tooling, and observability, balancing speed and quality. Descript HQ is in San Francisco, with remote and hybrid roles available.

Qualifications

• 8+ years building and operating production distributed systems.
• Experience using agents to multiply impact with AI.
• Reliability-focused background with incident management experience.
• Use of SLOs and error budgets as operating tools.
• Production Kubernetes and IaC in cloud environments.
• Owned architecture or migration end-to-end with lasting impact.
• Proactive in identifying unowned work and delivering with minimal guidance.
• Ability to build minimal repros, read logs, and write focused checks.

Responsibilities

• Own our platform: GCP, Kubernetes, Temporal, GPU fleet, and deploy/rollback machinery.
• Own the AI enablement substrate: training/inference pipelines, reliability and cost.
• Treat cost as an engineering constraint; implement metering and attribution.
• Ensure security boundaries: IAM, secrets, least privilege, supply-chain integrity.
• Make infrastructure legible: IaC, runbooks, in-repo docs, observability.
• Improve how the team learns and ships: hypotheses, instrumentation, incremental releases.
• Provide architectural direction, mentoring, and clear written communication.

Skills

Distributed systems

Agent-based AI

Reliability engineering

SLOs & error budgets

Kubernetes in production

Infrastructure-as-code

Architecture ownership

Proactive problem solving

Tools

Terraform / IaC

CI/CD pipelines

Cloud platforms (GCP/AWS)

GPU/ML infra tooling

Job description

Descript seeks an experienced platform engineer to own the foundation of our compute, deployment, reliability, CI/CD, and monorepo health for model training and serving. You will partner with engineers across teams to shape architecture and ensure robust systems.

You will work on on-call processes, tooling, and observability, balancing speed and quality. Descript HQ is in San Francisco, with remote and hybrid roles available.