A lightweight, cross-platform chaos engineering framework built in Rust for testing service resilience through controlled failure injection. Supports network latency, packet loss, CPU/memory pressure, and more on Windows, macOS, and Linux.

A middleware for LangChain agents that intentionally injects failures (exceptions) into tool and model calls to test agent resilience.

Graph Simulator lets you build and test network configurations with node redundancy and failure scenarios, providing insights into stability and necessary improvements. It logs disconnections and allows for multi-instance simulations to find optimal reliability.

AI-assisted lab management platform for DQA workflows (FastAPI + React + SQLite) with SOP/ISO 17025 automation and 140 tests (pytest 129 + Vitest 11).

End-to-end reliability testing for multi-step AI agent workflows. Chaos testing, cascade failure detection, E2E benchmarks. What DeepEval/Langfuse/LangSmith can't test. Works with LangChain, CrewAI, AutoGen, OpenAI. npm install agent-reliability

Public research artifacts, evaluation frameworks, prototype workflows, and technical documentation for LLM reliability, structured analysis, and applied AI systems.

Manual testing of a Productivity Suite application, validating core features for task management, collaboration, and reporting to ensure accuracy, usability, and reliability.

Measure LLM output consistency from the command line.

Replayable, inspectable evidence for reliability claims about stochastic systems. Falsification-first testing for LLMs — perturb inputs, preserve evidence, diff across model changes.