shakespear

pdfLLM is a completely open source, proof of concept RAG app.

A simple CPU only OCR for pdf/images/word/excel to markdown. With streamlit.

A simple streamlit app to play with qwen3-2b-VL to perform OCR. Dockerized set up, tested with 3060 12 GB.

A simple streamlit app, dockerized, to do OCR on documents. I'm lazy, idk.

The hunyuan 1B OCR model is pretty promising when it comes to OCR. It is lightweight, and very effective.

Turbo1Bit: Combining 1-bit LLM weights (Bonsai) with TurboQuant KV cache compression for maximum inference efficiency. 4.2x KV cache compression + 16x weight compression = ~10x total memory reduction.

This is the qwen3 4b instruct model. Being fine tuned for construction data.

Simple Accounting System for Contractors.

This is a pet project to use backblaze b2-s3 protocol with LanceDB for vector storage directly. Not too shabby.

This repo is only meant for MALV session usage and improvements.

exaPipeline is a sophisticated pipeline to process raw files (pdfs etc) into synthetic data for training. View exaOCR and exaPipelineDashboard

Extract Structured Data from PDFs, Word Docs and Images. Embeddable directly into your application, regardless of the stack.

JavaScript in-page GUI agent. Control web interfaces with natural language.

AI-assisted trading desk for short-term technical analysis on stocks & ETFs via Robinhood MCP. Deterministic Python engines score each asset on a three-pillar framework (Trend · Momentum · Macro-Sentiment) using EMA, RSI, MACD, TRIX & Bollinger Bands. The AI fetches data; the scripts compute; the human approves every order.

[DeepRead] This is the official implementation of the DeepRead paper.

A high-throughput and memory-efficient inference and serving engine for LLMs - Optimized for AMD gfx906 GPUs, e.g. Radeon VII / MI50 / MI60

Docker environment for developing, deploying, and running Frappe applications (ERPNext and custom apps) in production and development

Free and Open Source Enterprise Resource Planning (ERP)

A personalized bid mania for contractors. Easily log your project, add team members, and configure your email settings for automated notifications as you choose. Dockerized so you can easily deploy this anywhere.

A lightweight, lightning-fast, in-process vector database

This is a dockerized set up to run z.ai/GLM-OCR on your 3060 machine. It will require 575 drivers.

This is a production ready OCR Application with Streamlit for testing. Meant to be ran on an L40S 48 GB GPU. The model is Qwen3-4B-Instruct-FP8 for faster inference.

WIP testing.

This is strictly to be ran on a RTX Pro 6000 Blackwell. It is a FastAPI + Streamlit application meant to quickly test, and be scaled for batch OCR.

test

Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞

The world's most powerful open-source bio AI assistant - Access academic literature, clinical trials, drug labels, and more, all through natural conversation.

30 MCP tools for local AI agent swarms - runs entirely on qwen3:4b

vLLM for AMD gfx906 GPUs, e.g. Radeon VII / MI50 / MI60

Uses OpenAI Tokenizer + exaOCR and essentially gets the markdown format content and counts the token. Streamlit is there for interaction as always.

Generate clips from a video or youtube link

llama.cpp-gfx906

This streamlit app is a demo of exaPipeline as a demo. Feel free to recreate in whatever frontend you like.

This docker-compose.yml is going to pull the qwen3-4b-instruct-2507-awq and run it in vLLM. For ease of deployment on a AMD Instinct Mi50 (32GB)

high-performance graph database for real-time use cases

OCRmyPDF adds an OCR text layer to scanned PDF files, allowing them to be searched

A Python library for extracting structured information from unstructured text using LLMs with precise source grounding and interactive visualization.