Gen-AI-Hub

A curated collection of generative AI projects—LLM applications, retrieval-augmented generation (RAG) systems, fine-tuning experiments, and multimodal tools. Each project lives in its own folder with its own README, dependencies, and run instructions.

Author: @aniketpotabatti · License: MIT

---

Repository layout

Projects are grouped by theme. Open the folder for setup, architecture notes, and demos.

Folder	Focus
01-llm-apps/	Standalone LLM-powered apps (chat, summarization, codegen helpers)
02-rag-systems/	Document Q&A, hybrid search, and knowledge retrieval
03-fine-tuning/	Custom model training (LoRA / QLoRA)
04-evaluation-multimodal/	Vision, video, and VLM evaluation pipelines

Note: Some apps also appear at the repo root (PdfPal/, Dishcovery/, NutriGuide/) as earlier copies. Prefer the paths under the numbered folders above—they are the maintained locations.

---

01 — LLM applications

Project	Description	Stack (high level)
Automated Technical Writer	Paste code → production-style docs with streaming preview and Markdown/HTML export	Next.js, FastAPI, LangChain, Gemini
Blog to Podcasts Agent	Turn a blog URL into a short podcast script and MP3	Streamlit, Agno, Groq, Docling, gTTS
CodeSense	Real-time explain / review / suggest for code in the browser	React, Monaco, Express gateway, BYOK LLMs
Data Science Project Reviewer	Score a public GitHub ML repo across organization, metrics, deployment, and more	Streamlit, Gemini
Dishcovery	Ingredient-based recipe generation with dietary filters and PDF export	Streamlit, Gemini
End-to-End Youtube Video Summarizer	Summarize YouTube videos, Q&A on transcripts, export summaries	Streamlit, Gemini
OpsAgent	Analyze JSON system alerts → severity + summary with history	Streamlit, OpenAI / Gemini

---

02 — RAG & knowledge systems

Project	Description	Stack (high level)
AI Data Analyst	Ask questions in plain English over CSVs and PDFs; auto charts and insight memory	Streamlit, ChromaDB, Gemini
AI Research Explainer	PDF papers → study guides, flashcards, quizzes, ELI5	Streamlit, PyMuPDF, Gemini
PdfPal	Chat with PDFs and web pages; multi-provider LLM support	Streamlit, FAISS, sentence-transformers
SemanticSearchX	Modular RAG with hybrid retrieval, reranking, benchmarks, and observability	FastAPI, dense + sparse fusion, Prometheus
AI-powered finance report analyst	Placeholder — README and implementation coming soon	—
LLM Youtube Mixtape Creator	Placeholder — README and implementation coming soon	—

---

03 — Fine-tuning

Project	Description	Stack (high level)
Customer Support Bot	QLoRA fine-tune on Bitext support data; Gradio chat UI	Hugging Face Transformers, TRL, PEFT, Gradio

---

04 — Evaluation & multimodal

Project	Description	Stack (high level)
Image Caption Generator	BLIP + GPT-2 captions; API, CLI, and Streamlit UI	FastAPI, Streamlit, LoRA style control
NutriGuide	Food photo → calories, macros, health rating, diet tips	Streamlit, Gemini multimodal
Project Tagger	Structured product tags from image + text (Gemini, GPT-4o, Claude, LLaVA)	Pydantic validation, batch eval, Streamlit
VidLens AI	Video Q&A with Gemini plus live DuckDuckGo web grounding	Streamlit, Gemini 2.5

---

Getting started

1. Pick a project from the tables above and open its folder.
2. Read that project’s README.md for prerequisites, API keys, and run commands.
3. Use a virtual environment per project (python -m venv .venv or the stack’s documented setup).
4. Never commit secrets — use .env or session-only keys as each README describes.

Many apps need a Google Gemini, Groq, or OpenAI API key. Free tiers are documented in individual project READMEs where applicable.

---

Contributing

This hub is a portfolio monorepo. To extend it:

• Add new work under the matching numbered category (or propose a new category folder).
• Keep each project self-contained (requirements.txt, README, .gitignore).
• Follow existing naming and README patterns in sibling projects.

---

Security

See SECURITY.md for vulnerability reporting. Do not open issues with live API keys or private data.