[Hiring] MLOps Engineer @Akvelon

Role Description

This engagement is focused on building an internal AI platform that enables developers to ship AI-powered services efficiently. The objective is to improve DevEx and reduce time-to-market for AI features.

Location: Serbia (relocation support available), Croatia, Poland, Portugal

• Build and operate the AI platform infrastructure enabling developers to ship LLM-based services faster.
• Implement and maintain Kubernetes-based runtime environments (incl. AKS) for AI workloads.
• Manage infrastructure as code with Terraform (modules, environments, CI/CD automation).
• Support LLM workflows: RAG, agents, prompt experimentation, evaluations, and deployment patterns.
• Integrate and operate tooling such as Azure AI Foundry, LiteLLM, Langfuse, MLflow.
• Orchestrate pipelines using Kubeflow Pipelines and/or Argo Workflows (build, deploy, evaluate).
• Improve platform reliability and observability (monitoring, logging, tracing, cost/perf signals).
• Collaborate closely with developers to streamline DX (APIs, templates, docs, golden paths, automation).

Qualifications

• Strong hands-on experience with Kubernetes in production (preferably AKS).
• Solid Terraform expertise (IaC best practices, multi-env setups).
• Practical experience supporting ML/LLM workloads in a platform or DevOps/MLOps context.
• Proficiency in Python for automation, scripting, and supporting APIs/evaluation tooling.
• Understanding of CI/CD, release processes, and production-grade operations.
• Ability to work under tight timelines and deliver pragmatically.

Requirements

• Experience building internal developer platforms or “paved roads” for engineering teams.
• Familiarity with LLM evaluation frameworks, prompt testing workflows, and LLM observability.
• Exposure to RAG architectures, vector databases, and agentic patterns.
• Experience with Kubeflow, Argo, and ML lifecycle tooling.

Benefits

• Long-term B2B contract.
• Remote work from Croatia, Poland, Portugal, and Serbia.
• European working hours.
• Occasionally available for meetings up to 10:00 AM PST (US overlap).