Dependency Confusion: Supply Chain Attacks on Internal MLOps Tooling
A deep architectural breakdown of how PIP resolves namespaces and how attackers hijack internal MLOps tools via Dependency Confusion.
7 docs tagged with "mlops"
View all tagsDetecting Shadow AI: Auditing the Hugging Face Cache on Developer Endpoints
A practical guide to detecting Shadow AI infrastructure. Learn how to audit developer workstations and CI/CD runners for unvetted models hidden in the ~/.cache/huggingface directory.
O(1) Memory Fairness Evaluation for EU AI Act Article 10 Compliance
Learn how to calculate Demographic Parity and detect hidden proxy bias (Cramér's V) in 100GB+ training datasets without Out-Of-Memory (OOM) errors to comply with EU AI Act Article 10.
Securing Jupyter Notebooks: Mitigating 'Ghost Data' and Secret Leakage
Architectural strategies for securing Jupyter environments. Deep dive into nbformat JSON schemas, Shannon entropy for secret detection, and isolating execution states.
Securing the Dual-Track AI Supply Chain: PyPI and Hugging Face
A deep technical analysis of the dual-track AI supply chain: mitigating index resolution vulnerabilities in Python dependencies and identifying mathematical backdoors in ML artifacts.
YAML Deserialization Attacks: RCE via Configuration Parsing
A deep technical analysis of how PyYAML deserialization maps tags to Python object instantiation, leading to Remote Code Execution in MLOps configuration pipelines.
Zip Slip & Tar Bombs: RCE and DoS Attacks on AI Pipelines
How malicious archives target MLOps infrastructure, causing Out-Of-Memory (OOM) Denial of Service and Arbitrary File Overwrites (RCE), and how Veritensor mitigates them.