Model Online / On-Site
Duration
Modules 5 modules

Course Objective & Scope

Plaintext database passwords pasted into Jupyter notebooks, shared OpenAI/Anthropic API keys, unvetted Hugging Face checkpoints and over-privileged training pods with blanket S3 access are among the most common security exposures in enterprise AI platforms. This course brings Zero Trust and DevSecOps engineering directly to MLOps and GenAI infrastructure.

Identity federation for both interactive users and automated workloads (OIDC, Workload Identity/IRSA), dynamic secret injection with HashiCorp Vault and External Secrets Operator, Policy-as-Code on Kubernetes with OPA Gatekeeper and Kyverno, fine-grained data/vector access control and ML supply-chain security are covered hands-on. Participants design an auditable security and policy enforcement architecture for their AI platform.

Who Should Attend?

  • DevSecOps, cloud security and AI platform engineers securing data science and ML infrastructure
  • Systems architects designing authentication, authorisation and compliance controls for MLOps/LLMOps
  • Infrastructure leads supporting AI teams in regulated sectors (finance, healthcare, telecoms, public sector)
  • Prerequisite: Working knowledge of Kubernetes, containers, core IAM concepts and network security

Key Takeaways

  • Eliminate static credentials in notebooks and Git repositories using OIDC Workload Identity and short-lived tokens
  • Inject database credentials and LLM API keys dynamically at runtime with HashiCorp Vault and External Secrets Operator
  • Enforce Policy-as-Code with OPA/Kyverno to mandate signed images, approved model registries, non-root pods and GPU limits
  • Implement RBAC/ABAC row-, column- and document-level access controls across training datasets, Feature Stores and vector indexes
  • Build an air-gapped quarantine and scanning pipeline for external open-weight models, pickle files and container images

Curriculum

Threat Modelling AI Platforms and Workload Identity (IAM & OIDC)
  • Separating identities for interactive notebook users, CI/CD runners, batch training pods and autonomous agents
  • Replacing static cloud keys with OIDC Workload Identity (IRSA / Pod Identity) and least-privilege roles
Secrets Management: Vault, External Secrets and LLM API Keys
  • Runtime secret injection and automatic credential rotation with HashiCorp Vault and External Secrets Operator (ESO)
  • Blocking secret leaks in notebooks and Git with Gitleaks, and abstracting external LLM API keys behind a central AI Gateway
Policy-as-Code with OPA Gatekeeper and Kyverno
  • Kubernetes admission policies enforcing GPU resource quotas, non-root execution and trusted container registries
  • ML-specific guardrails: blocking production deployments of unapproved or unscanned model artefacts
Fine-Grained Access Control for Data, Registries and RAG Indexes
  • Column-level PII masking, row-level security and time-bound dataset grants across Lakehouses and Feature Stores
  • Document-level ACL enforcement at query time in RAG vector databases to prevent unauthorised context retrieval
Network Isolation, Egress Controls and ML Supply-Chain Security
  • Preventing data exfiltration from training environments with Kubernetes NetworkPolicies and egress firewalls
  • Quarantining external Hugging Face/PyPI artefacts, scanning weights with ModelScan, enforcing Safetensors and signing with Cosign

Other Courses in This Category

  • AI FinOps: Cost and GPU Capacity Management

    GPU and LLM token unit economics, cost visibility with DCGM/Kubecost/OpenCost, quota/priority scheduling with Kueue, MIG/fractional GPU sharing, Spot training and model routing, for AI platform and FinOps leads.

    MLOps & AI Platform Engineering Corporate
    Live Course
    5 modules
    View
  • Practical MLOps: End-to-End ML Pipelines

    Moving from Jupyter notebooks to production: modular Python code, DVC, MLflow, FastAPI packaging, Docker and GitHub Actions, for data scientists and engineers building a working end-to-end machine learning pipeline.

    MLOps & AI Platform Engineering Corporate
    Live Course
    5 modules
    View
  • MLOps – Fundamentals MLOps Fundamentals

    MLOps training covering machine learning model lifecycle management, containerization with Docker and Kubernetes, CI/CD automation with MLflow/Kubeflow, and model drift monitoring.

    MLOps & AI Platform Engineering Corporate, Individual
    Live Course
    6 modules
    View
  • Experiment Tracking and Hyperparameter Management in ML

    Systematic experiment design, hyperparameter optimisation with Optuna and Ray Tune, metric/artefact standards, reproducibility and comparative setup of tracking tools (MLflow, W&B, Neptune), for data science teams.

    MLOps & AI Platform Engineering Corporate
    Live Course
    5 modules
    View
  • MLOps – Advanced Level MLOps – Advanced Level

    Enterprise-grade MLOps covering model drift monitoring, A/B testing, canary deployments, automated retraining, CI/CD pipelines, Kafka/Spark Streaming, and Kubernetes orchestration.

    MLOps & AI Platform Engineering Corporate, Individual
    Live Course
    6 modules
    View
  • Model Tracking and Experiment Management with MLflow

    MLflow Tracking Server architecture, PostgreSQL/S3 setup, autologging, custom model packaging with pyfunc, MLflow Evaluate (classical ML and LLMs) and Model Registry integration, for teams standardising on open-source MLflow.

    MLOps & AI Platform Engineering Corporate
    Live Course
    5 modules
    View

Get in touch