We are looking for an experienced Senior DevOps / MLOps Engineer to join a high-impact project focused on the production deployment of an Anti-Churn Machine Learning system.
The goal of the project is to transition an analytical churn-prevention model from Proof of Concept (PoC) into a fully scalable, production-grade solution.
Key project pillars include:
- Prediction optimization – improving model performance using transactional and behavioral data (Oracle DWH)
- MLOps ecosystem development – full lifecycle automation (CI/CD) using Kubernetes, Apache Airflow, and MLflow
- Explainable AI (XAI) – implementation of SHAP to better understand customer behavior and decision factors
Key Responsibilities - Design and implement stable runtime environments based on Kubernetes (RKE2 / Rancher)
- Build and maintain end-to-end CI/CD pipelines using GitLab CI/CD and Nexus
- Orchestrate complex data pipelines and batch workloads in Apache Airflow (KubernetesPodOperator)
- Implement enterprise-grade security standards, including:
- HashiCorp Vault integration
- Secrets management
- RBAC policies
- Configure and maintain monitoring and alerting for ML systems (Prometheus, Grafana, ELK)
- Collaborate closely with Data Engineering and Data Science teams to optimize container resource usage
Must-Have Requirements
- 5+ years of experience in DevOps or System Engineering
- Expert-level knowledge of Kubernetes (administration, networking, storage), especially in on-premise environments
- Hands-on experience with MLOps automation and Apache Airflow orchestration
- Proven track record delivering secure solutions in regulated industries (FinTech, Banking, Insurance)
- Strong experience with CI/CD tools and containerization (Docker)
- Solid knowledge of MLflow (Tracking, Model Registry, Serving)
Nice to Have
- Experience with Infrastructure as Code tools (Terraform, Ansible)
- Familiarity with air-gapped environments (restricted internet access setups)
We hereby inform you that Inetum Polska sp. z o.o. has implemented an internal reporting (whistleblowing) procedure. The content of the procedure and the possibility to submit an internal report are available at:
https://inetum.whispli.com/speakup?locale=pl