SCIST x SCAICT Camp 2026 聯合寒訓 AI 系列課程 - 教材與原始碼

Production-ready semantic vector search for Django — searches across FK, M2M, and reverse relations by traversing your model graph. Pluggable backends: ChromaDB, FAISS, Qdrant.

🚗 Complete vehicle re-identification system: YOLO detection, OSNet embeddings, FAISS search. Session-based, no database, modern UI.

Text-to-image search engine for fashion runway photos using CLIP and FAISS.

AskMyPDF AI is a smart digital assistant that lets you "talk" to your PDF documents to get instant answers without reading through every page.

Full-stack RAG app — upload PDFs, ask questions. FAISS vector search + Groq LLM + Redis caching + per-user session isolation. Deployed with AWS Lightsail.

Document Q&A agent template with per-user JWT auth and a pluggable vector store — FAISS or ChromaDB, one env var to switch. SQLite-backed audit trail included. Intermediate tier of the Document Intel Agent Template family — for real users, not yet a full enterprise deployment.

Automatically turns research papers into fully structured Python projects. Using a multi-agent pipeline with RAG-based context retrieval. With Human-in-the-loop approval. Perfect for quickly prototyping implementations from academic papers.

SynthCerebrum is a fully offline, AI-powered assistant that reads and learns from all types of local files. Using advanced neural networks, embeddings, and RAG, it intelligently retrieves, synthesizes, and generates insights from your data, making your folder a brain-like knowledge hub.

RAG-based-on-PDF is a RAG system that lets users ask questions from one web page of VS code documentation which is scraped (given below) using FAISS, HuggingFace embeddings, and a OpenRouter Api via a FastAPI backend and Streamlit as Frontend.

A Vision-Language Model based multimodal retrieval system for semantic Image-to-Text and Text-to-Image search using OpenCLIP, FAISS indexing, and shared embedding space analysis.

VecFuzz is a lightweight approach to converts words into fixed-size numeric vectors for fast approximate string matching and typo-tolerant search.

A personalized, voice-first AI assistant for the visually impaired, combining real-time camera vision, speech recognition, and context memory.

A RAG chatbot that answers user questions using Wikipedia articles as context

Multi-document RAG assistant, upload PDFs, ask questions, summarize and compare

Cross-lingual (DE ↔ EN) hybrid document search over PDF/DOCX. Combines BM25, multilingual sentence embeddings (FAISS / Elasticsearch), proximity scoring, and optional DeepSeek (Ollama) for query expansion, re-ranking, and explanations. FastAPI backend + zero-dependency web UI.

The Medical RAG Chatbot is an AI-powered chatbot designed to provide accurate medical information using a Retrieval-Augmented Generation (RAG) approach. It combines large language models (LLMs), vector search, and PDF data to deliver precise, context-aware answers.

web_crawling_with_genaiAI is an AI-powered web crawling and content extraction system that collects website data, structures information intelligently, and enables smart search, summarization, and conversational Q&A using RAG, vector databases, and Generative AI

Deep Learning Image Based Fashion Recommendations

This repository contains a production-ready AI chatbot backend built with a strong emphasis on auditability, traceability, privacy, and compliance. It is designed for enterprise and regulated environments where every AI response must be explainable, reproducible, and verifiable.