MLOps Engineer (Istanbul / Ankara / Izmir)

About the Role

Huawei Turkey R&D Center is seeking a skilled MLOps Engineer to support our AI projects by building and maintaining scalable, reliable, and efficient MLOps infrastructures.

You will be responsible for managing the end-to-end lifecycle of machine learning models — from experimentation to deployment and monitoring — while working closely with data scientists and software engineers.

Key Responsibilities

• Automate ML model training, testing, deployment, and monitoring pipelines
• Design and maintain CI/CD workflows for ML systems
• Manage model versioning, experiment tracking, and model serving processes
• Collaborate with data science teams to optimize model performance in production
• Implement cloud-based MLOps solutions (Huawei Cloud or similar platforms)
• Ensure model observability, traceability, and security across all environments
• Establish and maintain comprehensive monitoring and logging systems (e.g., Prometheus, Grafana, ELK stack) to ensure system health and proactive issue resolution.
• Configure and optimize the performance of Nvidia GPUs for training and inference workloads. Experience with Huawei Ascend NPUs is a strong advantage and will be a key focus on the job.

Requirements

• Bachelor’s or Master’s degree in Computer Science, Electrical & Electronics Engineering, or related fields
• Minimum 2 years of experience in MLOps, DevOps, or ML infrastructure engineering
• Strong experience with Python, Bash, Docker, Kubernetes, Linux and Git
• Hands-on experience with CI/CD tools (Jenkins, GitLab CI, ArgoCD, etc.)
• Experience with ML pipeline tools such as MLflow, Kubeflow, or Airflow
• Familiarity with cloud platforms (Huawei Cloud, AWS, GCP, or Azure)
• Understanding of model monitoring and logging frameworks (Prometheus, Grafana, ELK Stack)

Preferred Qualifications

• Experience optimizing or deploying ML models in production
• Experience with API development or microservice architectures
• Familiarity with LLM, NLP or Computer Vision model deployment