instructor

OpenAI Python Library Tutorial by Datalumina®

Python library for the instruction and reliable validation of structured outputs (JSON) of Large Language Models (LLMs) with Ollama and Pydantic. -> Deterministic work with LLMs.

Extract structured data from any unstructured web page

Custom comfyui https://github.com/comfyanonymous/ComfyUI Nodes for interacting with Ollama https://ollama.com/ using the Instructor https://github.com/jxnl/instructor Library to provide structured output from your LLM

Type-safe structured data extraction from text using LLMs.

create pydantic models dynamically

Your personal contract AI assistant

Image Knowledge base / Idea / Problem Solve / Analysis

AI / ML Embeddings applied to the Steam Store.

Personal research knowledge base with an AI-scientist loop — deep-read papers into grounded bilingual analyses, incubate divergent questions, cross-analyze your library, connect your own experiments, iterate with round-memory advice.

Classify any text dataset with one config file. OpenAI, Anthropic, or Ollama — guaranteed structured outputs, unsupervised clustering, HTML reports.

Agentic Firewall v2 is a runtime security middleware and red-team benchmark harness designed to secure Model Context Protocol (MCP) applications. It acts as an intercepting proxy between an MCP client (AI agent)

A simple llm library

This project utilizes FastAPI and the Instructor (Python library) to ensure predictable, structured Large Language Model (LLM) responses, perfect for any web application backend.

Self-hosted AI tool catalog. Paste any URL — LLM extracts, classifies and indexes it automatically. Hybrid semantic search + AI chat.

Open-source evaluation suite that stress-tests LLM over-refusal on legitimate requests involving defeated, unjust, or absurd rules. 5 defeat families × 19 authority types = 95 scenario types.

Synthetic Q&A data pipeline for a Home DIY Repair assistant — LLM generation, quality gating, human + LLM-as-Judge labeling (6 dims), and iterative prompt correction proving a 92.7% failure-rate reduction.

Archbishop Mitty High School Religion Project Teaching

LLMs as production extraction infrastructure: rule-vs-LLM triage, validated structured outputs, cost-capped model routing, a measured precision/recall eval + vendor-drift detection, entity resolution, and an Airflow 3 ETL DAG. Runs fully offline.

Progressively extracts structured information from multi-turn user chats using pydantic schemas and LLMs