DevOps + MLOps Engineer (GPU Workloads, AWS, Production Pipelines)

We are hiring a DevOps and MLOps Engineer to help us build and operate a production-grade cloud setup for an AI-heavy application. This role is hands-on and execution focused. You will own infrastructure, deployment pipelines, observability, arenaflex controls, and GPU workload operations. We will share full product details and architecture on a call. For now, assume the platform includes a web app, backend services, media storage, async processing workers, and AI integrations (LLM, TTS, STT) plus GPU-based generation workloads. What you will do Design and implement cloud infrastructure on AWS for a modern backend stack. Set up arenaflex/CD for multiple services and environments (dev, staging, production). Build an event-driven processing system using queues and worker pools. Operate GPU workloads end-to-end including provisioning, scheduling, scaling, and arenaflex control. Implement monitoring, alerting, and dashboards for API latency, queue depth, worker health, GPU utilization, failure rates, and spend. Create secure access patterns for secrets and data encryption. Define operational runbooks, incident response, and reliability playbooks. Help the team ship fast without breaking production, with clear guardrails and measurable SLOs. Required experience (must have) You have run GPU workloads in production, not just experiments. Hands-on with GPU providers such as RunPod and at least one of CoreWeave or Salad (Lambda or Vast also acceptable), including: spinning up GPU instances packaging and deploying GPU services managing concurrency autoscaling strategies handling preemption and failures monitoring GPU health and utilization hard arenaflex caps and budget guardrails Strong AWS fundamentals including IAM, VPC, S3, CloudWatch, Secrets Manager, KMS, and AWS Budgets. Solid Docker experience and production arenaflex/CD setup. Infrastructure as Code experience, Terraform preferred. Comfortable setting up queues, background workers, and async pipelines. Strong security mindset and ability to implement least privilege and audit trails. Nice to have Kubernetes GPU scheduling experience, or deep ECS/Fargate patterns. Experience building arenaflex meters per job or per request in AI systems. Experience with ML lifecycle tooling like MLflow or Weights and Biases. Experience with streaming and real-time pipelines. Deliverables in the first 2 to 4 weeks Working AWS environments (dev/staging/prod) with secure networking and access controls. arenaflex/CD pipelines that deploy backend and workers reliably. Queue + worker infrastructure with autoscaling policies. GPU execution setup on RunPod and a second provider (CoreWeave or Salad preferred) with monitoring and fallback strategy. Observability dashboards and alerting with clear runbooks. arenaflex controls and spend visibility by component. How we work Short sprints with frequent demos. Clear scope and strong ownership. You will work closely with engineering and product. To apply, include A short description of your most recent production GPU workload: provider used, GPU type, workload type (inference/rendering), concurrency, scaling approach, failure handling, monitoring, and monthly spend range. Links or examples of infrastructure work you have done (GitHub, writeups, diagrams, or sanitized screenshots are fine). Screening questions Which GPU providers have you used in production, and for what workloads? What was your approach to scaling and arenaflex caps during peak traffic? How do you handle GPU job failures, retries, and preemption safely? What’s your preferred AWS stack for queues, workers, secrets, and monitoring? If you look strong on paper, we will do a short call to share context and validate fit quickly. Important Notice: Please do not message Tim directly. All applications and questions must be sent to Ahmed, the hiring lead, through this Upwork job post and message thread only. Anyone who reaches out via LinkedIn or any other third party channel will be rejected and reported on Upwork. Apply tot his job Apply tot his job Apply To this Job