LLM Expect introduces a code-first approach to LLM evaluation: run any Python function against a JSONL dataset and get automatic accuracy, schema fidelity, and judge-based scores. It’s a fast, transparent, zero-config tool for developers who want predictable, reliable LLM outputs without heavy infrastructure.

A fast, terminal-based tool for turning YouTube videos into clean, searchable transcripts using OpenAI Whisper. It downloads audio directly from any video, runs high-quality speech-to-text using your preferred Whisper model, and saves both timestamped and plain-text transcripts for easy review.

This tool streamlines your release workflow by pulling updates from JIRA, generating clean, structured release notes with OpenAI, and publishing them directly to Zendesk. It supports both CSV exports and live JIRA API calls.

This repository is a structured library of reusable model instructions built using Microsoft’s POML format. It organizes prompts, templates, functions, and multi-step chains into a single, consistent system that works across major AI frameworks like Azure AI Studio, Semantic Kernel, LangChain, and Prompt Flow.