AAIDC Module 1 Project: Ready Tensor RAG Assistant

🧠 Project Overview

This project implements a Retrieval-Augmented Generation (RAG) assistant as part of the Agentic AI Developer Certification (AAIDC) – Module 1. The assistant answers natural language questions using a custom knowledge base of simulated Ready Tensor publications, demonstrating core agentic capabilities: autonomous retrieval, context-aware generation, and goal-driven response synthesis.

The system is designed to be beginner-friendly, offline-capable, and fully self-contained — requiring no external APIs, no paid services, and no internet dependency beyond initial setup.

🛠️ Technical Implementation

Stack Used

• Framework: LangChain (for orchestration)
• Vector Database: FAISS (for fast similarity search)
• Embedding Model: sentence-transformers/all-MiniLM-L6-v2 (lightweight, CPU-friendly)
• LLM: facebook/opt-125m via Hugging Face Transformers (open-source, no API key)
• Document Source: 3 simulated Ready Tensor publications (covering RAG, Agentic AI, and vector databases)
• Interface: Jupyter Notebook (CLI-style Q&A)

Pipeline Flow

1. Document Ingestion: Sample publications saved as .txt files
2. Text Splitting: Recursive character splitter (200-char chunks)
3. Embedding & Storage: Chunks embedded and stored in FAISS
4. Retrieval: User query → top relevant chunks retrieved
5. Generation: LLM generates answer using retrieved context

✅ Key Features

• ✅ Autonomous RAG workflow (no manual intervention after query)
• ✅ Custom knowledge base (simulated Ready Tensor content)
• ✅ End-to-end local execution (runs on free Colab or personal machine)
• ✅ Sample queries & outputs included for validation
• ✅ Clean, documented code with setup instructions

▶️ Demo Output

When asked:

“What is RAG?”

The assistant responds:

“RAG combines retrieval from a knowledge base with LLM generation.”

When asked:

“How is memory implemented in agentic AI?”

Response:

“Memory is implemented via conversational buffer.”

When asked:

“Which vector databases are compared?”

Response:

“This publication compares FAISS, Chroma, and Pinecone.”

This demonstrates accurate retrieval and concise, context-aware generation.

📁 GitHub Repository

🔗 Repo Link: https://github.com/R786P/P-art

The repository includes:

• Rag_assistant.ipynb – Full RAG pipeline
• requirements.txt – Dependencies
• README.md – Setup and usage guide
• .gitignore – Excluded files

All code is clean, readable, and follows best practices for beginner-friendly AI projects.

🎯 Alignment with AAIDC Module 1 Requirements

🙏 Conclusion

This project marks my first step into agentic AI development. While simple, it demonstrates the foundational pattern of retrieval + reasoning that powers real-world AI assistants. I’m excited to build more complex agents with memory, tools, and planning in upcoming modules!

---

Submitted for AAIDC Module 1 Review Cycle – November 2025