Fast transformer inference for Ruby

Arabic diacriticizer from Interscript.

Detect credential stuffing, SQL injection, XSS, and other attacks using ML embeddings. Lightweight (~30MB model) with ~2ms inference time.

An AI-powered Ruby library that performs comprehensive on-device compromise detection for mobile applications. Features include root/jailbreak detection, emulator detection, hooking framework detection, application integrity checks, advanced network security analysis with certificate pinning and proxy detection, enterprise policy management, AI behavioral analysis, and RASP protection - all without requiring a backend.

OpenAI CLIP embeddings, uses ONNX models. Allows to create embeddings for images and text

A comprehensive forensic deepfake detection toolkit designed for law enforcement, military, and national security applications. Features AI/ML detection algorithms, forensic reporting, and court-admissible evidence generation with high accuracy deepfake detection for video, audio, and image files.

A Ruby port of FastEmbed - fast text embeddings using ONNX Runtime

Basic Ruby inference wrapper for the GLiNER2 ONNX model.

Denoises audio (Speech Enhancement) using GTCRN model

Easy to use some onnx models.

Determines file content types using original Magika's ONNX model

Detect NSFW images quickly using a pre-trained model ported from nsfwjs

Security pipelines for LLM applications using ONNX models from Hugging Face Hub. Detect prompt injections, jailbreaks, and PII leaks. Models are lazily downloaded and cached locally. Fast local inference (~10-20ms after initial load).

Ruby implementation of Google's SigLIP2 model for creating text and image embeddings. Uses ONNX models from HuggingFace onnx-community.