gaurdrail

Detect hallucinations in LLM responses. Verify every claim against source documents using hybrid STS + NLI. Works with LangChain, LlamaIndex, or any RAG pipeline. pip install longtracer

The immune system for AI coding agents

Protect your AI from Prompt Injection

A fail-closed runtime security layer for AI agents with prompt injection detection, tool-call protection, PII/secret sanitization, HITL controls, and output validation.

A beginner-friendly Responsible AI project exploring Foundation Models, prompt safety, toxicity behavior, AI guardrails, and fairness concepts using TinyLlama and Python on Kaggle.

Production-shaped agentic support helpdesk. LangGraph supervisor routes four specialist agents over pgvector RAG and three-tier memory, with code-level guardrails, human-in-the-loop writes, deterministic evals, and OpenTelemetry tracing.

Fine-tuned Llama-Prompt-Guard-2-86M classifier that blocks prompt injection attacks before they reach an LLM. +33 recall points over the base model on out-of-distribution attacks, ~20ms inference, served via FastAPI.