https://github.com/readytensor/rt-agentic-ai-cert-week3
Hands-on code and examples for Week 3 of the Ready Tensor Agentic AI Developer Certification Program - exploring LLM calls, multi-turn conversations, and secure system prompts.
https://github.com/readytensor/rt-agentic-ai-cert-week3
Last synced: 8 months ago
JSON representation
Hands-on code and examples for Week 3 of the Ready Tensor Agentic AI Developer Certification Program - exploring LLM calls, multi-turn conversations, and secure system prompts.
- Host: GitHub
- URL: https://github.com/readytensor/rt-agentic-ai-cert-week3
- Owner: readytensor
- License: other
- Created: 2025-06-01T05:45:24.000Z (about 1 year ago)
- Default Branch: main
- Last Pushed: 2025-07-21T00:07:26.000Z (11 months ago)
- Last Synced: 2025-07-21T02:22:09.401Z (11 months ago)
- Language: Python
- Size: 1.14 MB
- Stars: 1
- Watchers: 0
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Ready Tensor Agentic AI Certification - Week 3
This repository contains the lessons, practical code and exercises for **Week 3** of the [Agentic AI Developer Certification Program](https://app.readytensor.ai/publications/HrJ0xWtLzLNt) by Ready Tensor. Week 3 builds on the modular prompt engineering concepts introduced in Week 2 and adds practical examples of multi-turn conversations and robust system prompts.
## What You'll Learn
- Making your **first LLM calls** and grounding them in publication content
- Designing safe, consistent, and modular **system prompts**
- Managing **chat memory** in long conversations
- Adding **memory persistence** for session continuity
- Implementing a full **RAG pipeline** using vector databases and embeddings
## Lessons in This Repository
### 1. Your First LLM Calls: Getting Started with Groq & LangChain
Make basic LLM calls using LangChain and Groq. Learn to ground questions in documents and maintain context in multi-turn conversations.
### 2. System Prompts: Your AI’s Operating Manual
Design modular system prompts that define assistant behavior, tone, and safety boundaries. See how prompt structure impacts reliability and security.
### 3a. Memory Management Strategies: When Conversations Get Too Long
Explore how to manage growing conversation history using three strategies: stuffing, recent-only, and summarization — and analyze their trade-offs.
### 3b. Memory Persistence: Saving Conversations Across Sessions
Add memory persistence to your assistant so users can resume chats across sessions. Covers LangChain’s memory types and custom database implementations.
### 4. Building Your Research Assistant: A Step-by-Step RAG Implementation
Build a complete RAG system that semantically retrieves publication content and uses it to generate grounded, context-aware responses in real time.
## Repository Structure
```
rt-agentic-ai-cert-week3/
├── code/
│ ├── config/
│ │ ├── config.yaml # App config
│ │ └── prompt_config.yaml # Prompt configurations for system prompt examples
│ ├── paths.py # File path configurations
│ ├── prompt_builder.py # Modular prompt construction functions
│ ├── run_wk3_l1_example_1_2.py # Lesson 1: Basic LLM calls and publication grounding
│ ├── run_wk3_l1_example_3.py # Lesson 1: Interactive terminal chat example
│ ├── run_wk3_l2_sys_prompt_example.py # Lesson 2: System prompt examples and testing
│ ├── run_wk3_l3a_memory_strategies.py # Lesson 3A: Memory strategies comparison
│ ├── run_wk3_l4_vector_db_ingest.py # Lesson 4: Vector DB ingestion script
│ ├── run_wk3_l4_vector_db_rag.py # Lesson 4: Vector DB RAG example
│ └── utils.py # Utility functions
├── data/ # Sample publications for exercises
│ ├── 57Nhu0gMyonV.md
│ ├── ljGAbBceZbpv.md
│ ├── tum5RnE4A5W8.md
│ ├── yzN0OCQT7hUS-sample-questions.yaml # Sample questions used in lesson 3 exercise
│ └── yzN0OCQT7hUS.md
├── lessons/ # Lesson content and exercises
├── outputs/ # Generated prompts and LLM responses
├── .gitignore
├── LICENSE
├── README.md
└── requirements.txt
```
## Installation & Setup
1. **Clone the repository:**
```bash
git clone https://github.com/readytensor/rt-agentic-ai-cert-week3.git
cd rt-agentic-ai-cert-week3
```
2. **Install dependencies:**
```bash
pip install -r requirements.txt
```
3. **Set up your Groq API key:**
Create a `.env` file in the root directory and add your API key:
```
GROQ_API_KEY=your-api-key-here
```
You can get your API key from [Groq](https://console.groq.com/).
---
## Running the Lessons
This repository includes scripts to help you experiment with different types of interactions and prompt configurations.
### Lesson 1 — Basic LLM Calls & Multi-Turn Conversations
- **`run_wk3_l1_example_1_2.py`**
- **Example 1:** A simple LLM call answering a general question about VAEs.
- **Example 2:** The same question, but grounded in the publication content.
- **`run_wk3_l1_example_3.py`**
- An **interactive terminal-based chat** with the assistant, showing how multi-turn conversations work and how context is managed.
### Lesson 2 — System Prompts for Control & Safety
- **`run_wk3_l2_sys_prompt_example.py`**
- Tests system prompt configurations (basic and advanced) to see how they handle manipulative questions and maintain professional tone and safety.
### Lesson 3A — Memory Management Strategies
- **`run_wk3_l3a_memory_strategies.py`**
- Compares three memory management strategies: stuffing everything, trimming to recent messages, and summarizing conversation history.
- Simulates a long conversation using real questions, saving detailed results (Q&A pairs, token usage, and final prompts) in the `outputs/` directory.
- Includes an interactive mode for running a single strategy or a full comparison report.
### Lesson 4 — Vector Database & RAG Implementation
- **`run_wk3_l4_vector_db_ingest.py`**
- **Vector Database Ingestion:** Initializes a ChromaDB instance with persistent storage, chunks publications into smaller documents, embeds them using HuggingFace transformers, and stores them in the vector database for semantic search.
- **`run_wk3_l4_vector_db_rag.py`**
- **Retrieval-Augmented Generation (RAG):** Interactive terminal-based chat that retrieves relevant documents from the vector database based on user queries and generates contextual responses using retrieved content. Includes configurable similarity thresholds and result counts.
Each script saves outputs and transcripts in the `outputs/` directory for easy review and comparison.
---
## License
This project is licensed under the CC BY-NC-SA 4.0 License – see the [LICENSE](LICENSE) file for details.
---
## Contact
**Ready Tensor, Inc.**
- Email: contact at readytensor dot com
- Issues & Contributions: Open an issue or pull request on this repository
- Website: [Ready Tensor](https://readytensor.ai)