Echo

AI-reads-books-page-by-page. AI reads books: Page-by-Page PDF Knowledge Extractor & Summarizer. script performs an intelligent page-by-page analysis of PDF books, methodically extracting knowledge points and generating progressive summaries at specified intervals

GPT-4o-Research-assistant. The GPT-4o Research Assistant is a tool designed to leverage the power of GPT-4o in assisting with academic research. It searches for academic papers on ArXiv, identifies the most promising paper based on a given search term, downloads the paper, extracts its contents, and then summarizes it. The tool also suggests new search terms

deepdeek-r1-experiments. deep seek & o1 auto coders which write python code from a simple description and iteratively improvesit and fix errors

AI-book-maker-with-perplexity-search-grounding. uses gpt-4o and gpt-4-mini to write books on topics while researching with perplexity API

AI-Data-Scientist-scores-top-1-percent-on-Kaggle. An automated machine learning system that leverages O1 and Claude to iteratively develop, improve, and optimize ML solutions.

10-000-chatbot-prompts. A comprehensive collection of AI prompts with structured categories, subcategories, and searchable keywords. Each prompt includes detailed system messages and metadata for easy organization and discovery. 📚🔍

frankenthought-chat. uses all reasoning models in parallel and synthesizes an answer with o1. also has multi-chat where you can chat with any of them

deepseek-v3-iterative-auto-coder. an auto coder which automatically fixes errors and improves the code from simple user prompt

Real-time-voice-ultra-simple-template-with-function-calling. OpenAI real-time voice Fastapi template with function calling with maximum simplicity. comes with arxiv paper function as an example and full event capture

website-that-builds-itself. a website that builds itself

Arxiv-RAG-researcher. This tool allows you to search ArXiv for scientific papers, extract their content, embed and chunk the text, and ask questions about them using GPT-4o

o1-Auto-Teacher. An interactive AI-powered learning experience generator that creates comprehensive, multimedia educational content on any topic.

agentic-game-dev. 8+ agents work together to build a game in pygame

chat_with_data. Python

o1-divides-tasks-between-other-o1s. A powerful and simple asynchronous task management system that divides complex tasks into subtasks, processes them concurrently using o1 models, and synthesizes the results.

GPT-adversarial-defense. This Repo focuses on defending against 'adversarial prompts,' detecting and attempting to mitigate objectionable content in real time.

listen-to-deepseek-r1-thoughts. stream-of-consciousness experience of an AI's thinking process, complete with creative tangents and unexpected connections.

Swarm-library-tutorials. This includes 2 separate tutorial series for OpenAI swarm library each 10 files from basic to advanced

video-scene-splitter. splits videos into scenes with gpt-4o-mini and saves them separately

deepseek-r1-multi-stage-critical-thinking. implements an iterative critique system using multiple thinking AI models using openrouter

Multi-solution-o1-data-scientist. generates multiple parallel solutions. auto error corrects and iteratively improves

what-do-LLMs-think-the-most-important-list-maker-with-ranking. A specialized Python tool that collects, compares, and analyzes list-based responses from multiple Large Language Models (LLMs). Perfect for consensus analysis, trend identification, and comparative evaluation of different LLMs' responses to list-based queries.

Super-Simple-Fact-Checker-with-Perplexity-API. Super Simple Fact Checker with Perplexity Search

Single-batch-file-runs-4-script-content-creation-pipeline. Just run a single batch file and everything happens automatically: - No manual steps - No switching between programs - No waiting for one analysis to finish before starting another

deepseek-reasoning-explorer. Python

GPT-programming-tutor. Python

testing-deepseek-with-Groq-on-GPQA-diamond. testing deepseek with Groq on GPQA diamond

Tinder-like-Swipe-UI-for-finding-arxiv-papers. his is a Tinder-style paper discovery application for ArXiv research papers. Users can search for papers and swipe through them, saving interesting ones for later reference.

kinetic-low-poly-mechanisms. Cinematic Three.js low-poly mechanism and mathematical motion demos with gears, wave cores, nanomachines, transmissions, and Hopf-fiber blooms.

Dynamic-Language-Learning-Assistant-from-.txt-file-. you just write into the write_here.txt file and everything works dynamically: language and the text detection. it auto detects changes

Out-of-band-responses-template-for-OpenAI-real-time-voice-chat. includes 2 full fastapi webapp files 1: simple voice chat template 2: out-of-band responses voice chat template

Grokking-the-impossible-by-attempting-to-predict-the-digits-of-PI. An experimental machine learning project that attempts to predict digits of Pi using a neural network. This project explores whether a deep learning model can learn patterns in the sequence of Pi digits by encoding digit positions as binary combination representations, using spectral entropy regularization

how-accurate-are-LLM-facts. A sophisticated Python tool that generates interesting facts about any topic and automatically verifies them using dual AI systems. The tool uses OpenRouter for fact generation and Perplexity AI for fact verification, providing a reliable source of verified information.

Cellular-automata-templates-with-multi-core-and-GPU-optimizations. A high-performance, GPU-accelerated implementation of Conway's Game of Life with advanced visualization features. This project includes three different implementations, each optimized for different use cases.

JS-Package-recommender. This is a web application that helps developers find suitable JavaScript libraries for their projects using voice input and AI recommendations. Users can speak their project requirements into their browser's microphone, and the application transcribes the speech in real-time. The transcribed text is then sent to a FastAPI backend server,

marked. A markdown parser and compiler. Built for speed.

neural-net-learns-without-backprop. This project implements a basic neural network from scratch that learns through random weight adjustments rather than traditional backpropagation. The network attempts to learn a simple addition operation.

swarm-seeds. Tested, reproducible orchestration patterns for building more capable AI-agent swarms.