Lester Oliver

ai-pdf-ppt-generator-openai. A fun project where I use the power of AI to analyze a PDF. The AI extracts key information based on the user's instructions and selections (see the UI demo). The user then gets a second screen to edit the slides before downloading the final PPT. Simple, fast, and powered by AI to make creating presentations a breeze!

microsoft-markitdown-streamlit-ui. A document analysis tool built with Streamlit and Microsoft MarkItDown. Extract and analyze content from multiple document formats with optional GPT-4o enhancement.

docling-pdf-processor. PDF intelligence platform combining IBM Docling for document processing, LlamaIndex for data structuring, and Streamlit for a powerful UI. Extract, analyze, and interact with documents using AI capabilities.

contextual-chunking-graphpowered-rag. An advanced retrieval system that combines semantic vector search with token-based search, using contextual chunking and knowledge graphs to improve accuracy. It ensures precise query responses with LLM-based answer checking and graph visualization.

contextual-doc-retrieval-opneai-reranker. Contextual Doc Retrieval is a Python-based system leveraging OpenAI GPT-4o and Cohere for re-ranking and query expansion, combined with BM25 for accurate document retrieval. It parses PDFs, chunks content contextually, and enhances search precision with AI-powered contextual understanding and re-ranking.

openai-rfp-response-analyzer. RFP-Response Analyzer is a Flask web app that uses AI to analyze and compare RFP documents with their responses, providing insights, gap analysis, and interactive chat for quick queries.

ai-dataset-generator. I got tired of manually creating training datasets, so I built this. Transform your PDFs/docs into fine-tuning data automatically.

rfpanalyzer-gemini2-flashthinking. Streamlit application that helps users analyze RFP's using the latest Gemini 2.0 Flash Experimental LLM.

llamaindex-agentworkflow-browse-agent. Automate web interactions using LLMs and Llama-Index AgentWorkflow

agno-deepknowledge. An implementation of an iterative knowledge base search agent using Agno agent framework, inspired by Ashpreet DeepKnowledge concept using Agno

google-gemini-pdf-table-extractor. This experimental tool leverages Google's Gemini 2.5 Flash Preview model to parse complex tables from PDF documents and convert them into clean HTML that preserves the exact layout, structure, and data.

deepseek-r1-groq-rag. A Retrieval-Augmented Generation (RAG) system running DeepSeek R1 Distill LLama 70B model using Groq's fast inference API.

phi-agent-showcase. A showcase of Phi Agent capabilities featuring a Streamlit-powered Research and Finance AI assistant built with Meta Llama 3.3 70B Instruct Turbo

pdfgenius-gemini-2.0-flash-exp. A Python-based RAG application that uses Google's Gemini-2.0-flash-exp API to analyze PDF documents and answer questions about their content. Turn any PDF into an interactive knowledge base with advanced vision-language capabilities.

pdf-analysis-claude-sonnet. This app uses Claude 3.5 Sonnet to analyze PDFs. Process single PDFs Batch process multiple PDFs Cache results for faster repeated queries

openai-swam-simple-rag. This repository demonstrates a simple OpenAI Swarm-based system for multi-agent orchestration with Retrieval-Augmented Generation (RAG). It handles tasks like summarization, sentiment analysis, keyword extraction, and document search using FAISS and OpenAI models, showcasing the power of collaborative agents.

pdf-analyzer-with-page-citations. An experimental implementation for PDF document analysis with accurate citation capabilities using the Agno Agent Framework.

espclaw-esp32-s3-n16r8. ESP-Claw AI agent firmware for ESP32-S3 N16R8 (16MB flash + 8MB PSRAM) — pre-built binaries, no ESP-IDF required

economic-analyst-ai-agent. a lightweight AI-powered Economic Analyst Agent built using the Agno-AI framework. The agent leverages Ollama (Llama 3.2) with a Retrieval-Augmented Generation (RAG) system, combining structured knowledge from financial reports with real-time web searches.Agno AI Economic Analyst Agent

agno-reasoning-agents-local. A powerful AI assistant that thinks, researches, and explains complex topics—all on your local machine. Powered by Agno's multi-agent framework with built-in reasoning, it analyzes questions, evaluates evidence, and draws logical conclusions. Uses Qwen3 through Ollama for complete privacy with no data sent to external servers.

video-analysis-gemini2flashexp. A web application powered by Google's latest Gemini 2.0 flash exp model that helps you analyze videos. Simply upload a video and chat with the AI to get detailed insights about its content.

gemini-vision-langextract-streamlit. Google Gemini 2.5 Pro (Vision + thinking) + google/langextract with a tiny Streamlit UI and a CLI. Point it at PDFs (scans included), DOCX, or HTML and get grounded clause/entity extractions with a quick HTML viz and CSV export. Local dev first. MIT.

late-chunking-embeddings. Late Chunking Embeddings: A Python project that demonstrates late chunking and embedding comparisons using transformers. It compares the effectiveness of traditional and late chunking methods for generating embeddings and measuring similarity.

pdf-markdown-visionparse. A simple Streamlit UI wrapper for vision-parse to convert PDFs to Markdown using Multimodel AI

ai-piano-player. The AI Piano Performer is an interactive web application that generates piano music using AI based on natural language prompts. Simply describe the kind of music you want to hear, and watch as the AI creates a unique piano composition tailored to your request.

ai-model-compare-aisuit. Streamlit UI to test and compare different AI models using Andrew Ng's AISuite lib. It's basically a playground where you can throw prompts at multiple models and see how they respond side by side.

ai-multiagent-discussion. An experimental AI implementation exploring multi-agent conversations using ControlFlow and Firecrawl.

license-vision-analyzer. A Streamlit app using Google Gemini Flash Thinking to intelligently extract and analyze user-specified entities from driver's licenses through natural language prompts and delivering structured results

groq-doc-vision-chat. Document analysis assistant using Groq's LLama models for vision processing and interactive Q&A, built with Streamlit.

rag-chonkie-llamaparse-openai. A practical implementation showcasing the power of late chunking strategy using Chonkie, combined with LlamaParse, OpenAI and LangChain.

google-deplot-streamlit. A streamlined web app that instantly converts chart images into downloadable CSV data using Google Deplot AI-powered extraction.

vision-analyzer-bench. A streamlit-based tool for benchmarking and comparing different Vision LLMs (Ollama, Groq, Gemini).

openai_o1_math. Python

resume-analyzer. An AI-powered application that analyzes resumes against job descriptions using LangChain, OpenAI, and LlamaParse.

wellness-coach-openai-realtime. a simple application build with the openai realtime api

ai-icon-generator-openai-dalle3. Icon generator using OpenAI Dall-E3

poofmac. 💨 Local AI-powered Mac disk cleaner. Runs on Ollama — no subscription, no data leaves your machine.