Ameriabank CJSC
Armenia
Industry:
Finance/Banking/Insurance
Number of Employees:
500-1500
Type:
LLC/OJSC/CJSC
MLOps Engineer_MLOps platform Squad
Full time
Yerevan
20 November 2026
Employment term Permanent
Category Data Science/Data Engineering
Job description:
Building and operating a modern on-premise DataOps and Machine Learning platform for Data Engineering and Data Science teams. The platform is based on Kubernetes and supports scalable, high-availability workloads across data processing, analytics, and ML systems. The role combines DevOps engineering, modern data lakehouse engineering, and MLOps support
Job responsibilities
- Design, deploy, and operate production Kubernetes clusters (on-premises) with focus on high availability, scalability, and zero-downtime deployments
- Manage Helm charts and ArgoCD for GitOps-based continuous delivery of platform and application workloads
- Build and maintain CI/CD pipelines in Azure DevOps for automated testing and deployment across data and ML services
- Architect and maintain a modern data lake using open table formats (Iceberg) on on-premises and hybrid storage
- Build, schedule, and monitor scalable batch and streaming pipelines using Apache Airflow, Apache Spark, and Apache Flink
- Deploy and operate Trino and ClickHouse as query engines for analytics over the data lake
- Deploy and operate JupyterHub and Dagster on Kubernetes for ML experimentation and workflow orchestration
- Manage the Feature Store (Feast / PostgreSQL) and Parquet-based data storage for ML workloads
- Work closely with Data Science and Data Engineering teams to productionize ML workflows and onboard new data sources
Required qualifications
- Hands-on experience designing and managing Kubernetes clusters from scratch (on-premises preferred)
- Hands-on experience with Helm and ArgoCD for GitOps-based application delivery in Kubernetes environments
- Experience building CI/CD pipelines with Azure DevOps
- Experience with Apache Airflow, Apache Spark, and Apache Flink for pipeline orchestration and processing
- Experience with Trino or ClickHouse as distributed query engines
- Experience with dbt for data transformation and modeling
- Familiarity with modern open table formats (Iceberg)
- Familiarity with Dagster, JupyterHub, Feast, or Parquet-based data lake patterns will be a plus
