Senior DevOps with Python 8648 G

Join Akvelon — build products used by millions!

Akvelon is an IT company with 20+ years of experience and 1,200+ engineers across 15+ locations worldwide.
We work with both well-known global tech companies, including Microsoft, Facebook, Airbnb, Dropbox, and Pinterest, and with growing startups.
The projects cover a variety of engineering areas, ranging from cloud solutions and AI/ML systems to big data, web, and mobile applications.
Collaboration typically takes place in a remote, distributed environment across different locations and time zones. We value ownership, clear communication, and accountability for agreed deliverables throughout each project.

About the role
The project expands an advanced Kubernetes-based inference benchmarking framework to provide end-to-end performance measurements, cross-cloud comparisons, and updated model test scenarios.
It includes automated benchmarking for startup latency and scale-out behavior across multiple model families and hardware configurations, integrates dashboards and reporting, and supports next-generation autoscaling, scheduling, storage, and multi-host workloads to ensure reliable measurement of real-world inference performance.

Requirements

• Advanced Kubernetes expertise including deep understanding of pod lifecycle, deployments, services, autoscaling and troubleshooting, with hands-on experience in GKE
• Experience with Python, including automation, scripting and infrastructure skills
• Experience with observability, monitoring, logging, tracing and performance benchmarking
• Practical experience with GCP services including GKE, GCS and Cloud Monitoring / Logging
• MLOps experience including deploying and operating ML models, working with vLLM or similar frameworks and managing GPU workloads in Kubernetes
• Solid understanding of autoscaling with HPA, Metrics Server and basic knowledge of Cluster Autoscaler
• Experience designing and maintaining CI/CD pipelines using GitHub Actions
• Strong Python skills for automation, scripting and infrastructure tasks

Nice to Have

• Knowledge of AWS and Azure cloud services

Responsibilities

• Provide consistent, platform-wide performance signals across all inference workloads and teams, ensuring clear visibility into system efficiency and bottlenecks
• Deliver standardized cross-cloud benchmarking across major Kubernetes providers to ensure reliable performance comparisons
• Support leadership reporting through monthly cloud-wide performance results that enable accurate insights and data-driven decision making
• Enable teams to validate new features including autoscaling, scheduling, storage, node provisioning, vLLM optimizations, and accelerator support using unified benchmarking frameworks aligned with organizational OKRs
• Extend the benchmarking framework to cover startup latency and scale-out behavior for multiple model families and hardware configurations
• Integrate automated benchmarking APIs, dashboards, and reporting pipelines to streamline performance evaluation
• Collaborate with engineering teams to maintain reusable inference components while ensuring accurate scheduling, infrastructure provisioning, and reporting

Availability expectations

• Flexibility, ability to work with overlap until 11 AM PST