MLOps Infrastructure Engineer

We’re looking for an infrastructure engineer with strong Go and Kubernetes skills and hands-on experience running AI models to help build and operate our platform.
In this role, you’ll work at the intersection of backend engineering, cloud infrastructure, and AI model deployment, turning compute resources into reliable services for AI workloads.

What you’ll do:

• Develop backend services and infrastructure automation in Go.
• Build and operate Kubernetes infrastructure for GPU-based training and inference workloads.
• Design workflows for deploying, scaling, updating, and removing model endpoints.
• Package and deploy different types of AI models, including LLMs, speech models, embedding models, and vision models.
• Integrate model-serving frameworks and optimize GPU utilization, inference latency, and throughput.
• Automate model and dataset access, container setup, storage provisioning, and workload scheduling.
• Implement monitoring, logging, and alerting for infrastructure and model-serving services.
• Troubleshoot issues across applications, containers, networking, storage, and GPU workloads.
• Collaborate with backend and AI engineers to improve platform reliability and developer experience.

What we’re looking for:

• Strong Go programming skills, with experience building backend services, APIs, or infrastructure tools.
• Hands-on experience deploying and troubleshooting workloads on Kubernetes.
• Solid understanding of Linux, Docker, networking, and storage.
• Practical experience running and serving AI models, including managing dependencies, model weights, and GPU memory requirements.
• Familiarity with the model lifecycle: training, packaging, deployment, versioning, and inference.
• Working knowledge of Python and AI frameworks such as PyTorch or TensorFlow.
• Experience with Git, CI/CD pipelines, and deployment automation.
• Ability to investigate production issues and take ownership of reliable solutions.

Nice to have:

• Experience with serving tools such as vLLM, NVIDIA Triton Inference Server, KServe, or Ray Serve.
• Experience with NVIDIA drivers, CUDA, GPU Operator, and GPU scheduling.
• Familiarity with quantization, batching, inference caching, and multi-GPU inference.
• Experience with Helm, Kubernetes operators, or custom controllers.
• Familiarity with Prometheus, Grafana, and distributed tracing.
• Experience with S3-compatible object storage, model registries, and artifact management.
• Experience building multi-tenant platforms with resource quotas and workload isolation.

What success looks like
You’ll help make it straightforward for Afran users to run training jobs and deploy reliable model endpoints, while improving resource efficiency, observability, and day-to-day operations.