Skill UI

A comprehensive guide to managing training data and model artifacts securely on cloud GPU instances (e.g., Vast.ai). It covers best practices for data transfer (SCP, compressed, cloud staging), implementing AES-256 encryption, managing checkpoints to S3, and performing secure cleanup before instance destruction, ensuring compliance and data integrity.

This guide provides a robust implementation for handling asynchronous webhooks from ElevenLabs. It covers the critical steps of receiving POST notifications (such as transcription completion, post-call audio, and failure events), performing secure HMAC-SHA256 signature verification, and routing the event data to specific processing handlers. Ideal for integrating ElevenLabs into conversational AI pipelines, ensuring data integrity and reliable background processing.

This skill provides comprehensive tools to detect data poisoning and model backdooring attacks throughout the machine learning lifecycle. It checks data provenance, detects label-flip anomalies (using Cleanlab), and analyzes model activations and spectra (using ART) to identify malicious triggers or compromised weights, ensuring ML supply chain integrity.