QED42 ORG

A demo app for omniauth with authlogic

Headway - Selenium Maven TestNG POM Data Driven Framework

No description.

This script scans a directory for Twig files, extracts variable definitions and options from comments, and generates the corresponding .component.yml files required for SDC components.

A Streamlit demo of Hugging Face's SmolAgents library, showcasing a CodeAgent that can reason over natural language queries and execute code. The agent is equipped with DuckDuckGo search so it can pull in live web results as part of its responses.

AI Accessibility Checker is a Python CLI and GitHub Action that scans frontend code for WCAG compliance (2.0–2.2, A/AA/AAA). It uses OpenAI to detect issues and provide clear, AI-driven fix suggestions for HTML, CSS, JSX, and more.

A polymer element to evenly distribute the photos across rows or columns, making the most of the space provided.

No description.

This is a polymer element to indicate the amount of article still left for reading an a page.

An alternate multimodal RAG approach that converts extracted text and images from PDFs into LLM-generated summaries before embedding them. Text is summarized with Groq's Llama-3.1-8b-instant and images/tables with GPT-4o-mini, with everything stored in a Chroma vector store for retrieval.

A Jupyter notebook demonstrating how to use the Qwen2-VL vision-language model to understand and chat about video content. It extracts frames from a video and supports an interactive, multilingual conversational interface for querying what's happening in it.

An AI News Blog Generator that fetches the latest AI news via the GNews API or accepts a custom article link. It then scrapes the content and uses GPT-4o-mini to write a structured, polished blog post automatically.

Drupal 6 theme

Drupal 6 theme suitable for professional bloggers

Simple like/dislike rails plugin

Port of iNove theme to drupal 7, uses blueprint css framework for ease of development

Drupal module to demo migration from a simple rails app, uses migrate api version 2.

Rails part of demo rails to drupal migration.

A polymer element to use zoomove jquery plugin.

No description.

No description.

A FastAPI-based voice assistant application with real-time voice activity detection and WebSocket-based audio streaming. It transcribes speech using Groq's Whisper model, generates AI responses, and converts them back to speech via OpenAI TTS.

A multi-agent content pipeline built on OpenAI's Swarm library, where a researcher agent gathers information via the Tavily API and a blogger agent turns it into polished content. A user-interface agent coordinates the handoff between research and writing in an interactive demo loop.

An AI-powered counterpart to the script-based HTML-to-Twig converter, using OpenAI's API for more intelligent content transformation. It converts HTML into Twig templates with auto-generated loops and conditionals, and converts CSS into SCSS with variable substitution and inline explanations.

Demonstrates how to use Weights & Biases' Weave extension to evaluate and compare LLMs on contextual relevance for question answering. It implements a full RAG evaluation pipeline with custom scorers and a dashboard for visualizing model performance side by side.

A Streamlit web interface built on Andrew Ng's AISuite library for comparing responses from multiple LLM providers side by side. It currently supports OpenAI GPT-4o, Anthropic Claude 3.5 Sonnet, and Groq's LLaMA 3, with adjustable temperature and system prompts per request.

A high-performance RAG application for chatting with documents, built on Llama 3.3 served through SambaNova's lightning-fast inference engine. It uses LlamaIndex for orchestration and Qdrant as the vector store, with a Streamlit front end for uploading documents and querying them.

A scraper that automates extracting scholarship listings from target websites using Selenium. Google's Gemini AI then structures the raw scraped data, which is exported into a clean Excel file for review.

A Streamlit shopping assistant that uses Google's Gemini AI to analyze an uploaded product image. Based on user-set preferences like color, purpose, and budget, it returns tailored shopping recommendations with direct product links.

A browser extension that runs a quantized Phi model entirely on-device using Web-LLM/WebGPU, built as a take-home assignment exploring local AI in the browser. It supports document summarization and basic code generation fully offline, prioritizing privacy and snappy client-side performance over cloud inference.

A notebook comparing six different text-chunking methods for RAG pipelines, ranging from manual fixed-size splitting to LLM-driven agentic chunking. Each method is paired with guidance on which use case it suits best, from simple document processing to deep knowledge retrieval.

A CrewAI-orchestrated RAG workflow for answering questions about research papers, combining a PDF vectorstore with live web search. Router, retriever, grader, and hallucination-filtering agents work together to ensure the final answer is relevant and factually grounded.

A set of scripts that analyze a UI screenshot using the Google GenAI API and break down the visible web components into a structured format. Output can be generated as either cleaned JSON or YAML depending on which script variant you run.

A Streamlit-powered data visualization agent that lets you upload a dataset and ask questions about it in plain English. It auto-selects appropriate chart types and runs the analysis through a sandboxed code-execution environment, drawing on multiple LLMs for different levels of analytical depth.

A chat-based assistant that lets you control your laptop through natural language instead of digging through settings manually. It can check system specs, manage files and processes, and run Docker or Git commands, while requiring manual confirmation for any sudo operations.

A Streamlit app that runs traditional vector-based RAG and a Neo4j knowledge-graph-enhanced RAG pipeline side by side on the same documents and questions. Each answer is shown with its response time so you can directly compare retrieval quality, reasoning depth, and latency between the two approaches.

A collection of small-scale ML models, including DistilBERT, Logistic Regression, and a Random Forest classifier, trained on dummy datasets for exploratory query classification. Built to demonstrate how each model handles legal-aid query categorization tasks of varying complexity.

No description.

An AngularJS wrapper for select2

AngularUI wrapper for TinyMCE

Admin menu and features packages for mean

meanio core module

Mean Io Repository

PM2 Weekly Task Scheduler

No description.

Demo module for the lazybuilder concept

No description.

No description.

Directive for JScrollPane

Drupal Campus Ambassador Program

Git repo for Drupal Camp Website

Fork of tinymce with fixes

No description.

A robust, reusable Cypress Page Object Model framework for scalable end-to-end testing

No description.

No description.

No description.

No description.

No description.

No description.

No description.

No description.

Milvus Docker Compose

DEMO: Dummy Data Processing with Ray and FastAPI

This demonstrates automation tests for eCommerce Journey using Headway framework

Package files of Masonry layout to build in Acquia Site Studio.

No description.

No description.

No description.