This image shows an automated machine learning operations (MLOps) pipeline on Oracle Cloud Infrastructure (OCI) that links internet participants with managed DevOps and data science services for one availability domain.

• Internet participants: Customers send inference requests to Oracle Cloud Infrastructure Web Application Firewall. Code developers push code to a code repository that is used by Oracle Cloud Infrastructure DevOps; data scientists and ML developers interface with Oracle Cloud Infrastructure Data Science.
• OCI region
  • Compartment: Provides governance and shared policies for the MLOps workloads in this region.
  • Availability domain: Supplies isolated compute capacity that operates the MLOps networking and endpoints.
  • Oracle Services Network provides:
    • Oracle Cloud Infrastructure Identity and Access Management
    • Oracle Cloud Infrastructure Vault
    • Oracle Cloud Infrastructure Logging
    • Oracle Cloud Infrastructure Monitoring
    • Oracle Cloud Infrastructure Notifications
    • Oracle Cloud Infrastructure Object Storage
  • Virtual cloud network (VCN)
    • Subnet A: The public subnet runs OCI WAF, Oracle Cloud Infrastructure Load Balancer, an internet gateway, and a network address translation (NAT) gateway; OCI WAF filters requests and passes approved calls to OCI Load Balancing, which distributes them by using the VCN; the NAT gateway lets private workloads reach Oracle Cloud Infrastructure Registry without exposure.
    • Subnet B: The private subnet runs OCI Kubernetes Engine model endpoints, MLFlow endpoints, and Data Science jobs; OCI DevOps pulls from the developer code repository, builds container images and stores them in OCI Registry; Data Science tracks experiments and promotes trained models to the endpoints by using OCI DevOps pipelines; Data Science models and workload results are stored in OCI Object Storage.