Course Objective & Scope
While application code typically passes through strict CI/CD and DevSecOps gates, data science environments often rely on over-privileged Jupyter notebooks, unauthenticated MLflow or Kubeflow dashboards and unsigned model artefacts promoted straight to production. This course adapts DevSecOps practices to the machine learning lifecycle (MLSecOps) to establish an unbroken chain of trust from data ingestion to model serving.
Participants configure hardened MLflow/Kubeflow setups, automated CI/CD scanning with ModelScan, cryptographic model and container signing with Sigstore/Cosign, and least-privilege inference deployment on Kubernetes and KServe.
Who Should Attend?
- MLOps, LLMOps, DevOps and platform engineers
- DevSecOps, cloud security and infrastructure security teams
- Machine learning engineers deploying models to production
- Prerequisite: familiarity with Docker, Kubernetes, CI/CD pipelines and core MLOps tooling
Key Takeaways
- Lock down privilege and credential exposures across Jupyter notebooks, feature stores, training clusters and model registries
- Integrate model serialisation scanning (ModelScan), SCA and container checks into ML CI/CD pipelines
- Cryptographically sign model weights and build SLSA provenance attestations using Sigstore/Cosign
- Enforce admission controls so KServe, Triton or vLLM pods load only verified, signed models
- Configure network policies, secrets management and runtime monitoring on multi-tenant GPU Kubernetes clusters
Curriculum
The MLOps Attack Surface and MLSecOps Architecture
- Exposures across notebooks, object storage, experiment trackers (MLflow) and orchestrators (Kubeflow/Airflow)
- Lessons from real ML platform breaches: SSRF, remote code execution and leaked cloud credentials
Securing Data Science Workspaces and Training Clusters
- JupyterHub isolation, ephemeral workload identities (IRSA/Workload Identity) and secrets management with Vault
- Egress controls on training pods and secure feature store patterns that prevent raw production data sprawl
Model Scanning, Signing and Provenance in CI/CD
- Embedding ModelScan, Trivy and CycloneDX AI-BOM generation into CI/CD pipelines
- Signing model weights and generating SLSA provenance attestations with Sigstore Cosign and OpenSSF Model Signing
Model Registry Governance and Promotion Gates
- RBAC, approval workflows and immutable artefact versioning in MLflow and cloud model registries
- Automated security, robustness and performance gates before promoting models to production
Hardening Kubernetes and KServe Inference Infrastructure
- Admission control with Kyverno/OPA Gatekeeper to allow only signed containers and model artefacts
- GPU isolation considerations (MIG vs time-slicing), mTLS service mesh and Falco runtime monitoring
Other Courses in This Category
-
Covers the full OWASP Top 10 for LLM Applications—including direct and indirect prompt injection, improper output handling, unbounded consumption and sensitive data disclosure—from a secure software development (SSDLC) perspective.
- Live Course
- 5 modules
-
Hands-on training on conducting end-to-end penetration tests against classical ML, deep learning and LLM APIs using MITRE ATLAS, Counterfit, ART, garak and PyRIT, and producing actionable pentest reports.
- Live Course
- 5 modules
-
Hands-on training on applying machine learning and LLMs across cybersecurity—from phishing and malware detection to vulnerability prioritisation, cyber threat intelligence (CTI) analysis and SecOps assistants.
- Live Course
- 5 modules
-
Teaches how to detect malware, ransomware, C2 traffic and multi-stage attack chains across EDR, NDR and cloud telemetry using behavioural machine learning, sequence models and graph analytics.
- Live Course
- 5 modules
-
Teaches how to build unsupervised and semi-supervised anomaly detection pipelines on logs, network traffic and user behaviour (UEBA) using Isolation Forests, Autoencoders, time-series and graph methods while minimising false positives.
- Live Course
- 5 modules
-
Teaches how to integrate LLM assistants, natural-language-to-KQL/SPL query generation, automated alert enrichment, triage agents and incident summarisation safely into SIEM, SOAR and modern SOC operations.
- Live Course
- 5 modules
Get in touch
