/ cloud /

Developing Intelligent Systems with LLMs and RAG

Build web-based applications that integrate LLMs.

Workshop Information

Title: Developing a Web-based Q&A system using a RAG-based approach to LLMs
Instructor: Sujee Maniyam, Node 51
Time: October 9, 2024; 11 a.m. – 5 p.m. ET
Duration: 6 hours
Cost: $300 for NET+ AWS and GCP Subscribers; $400 for Internet2 members; $500 for non-members
Delivery: Virtual via Zoom
Registration: Register

Workshop Overview

Large Language Models (LLMs) are reshaping how we approach knowledge discovery, data processing, and personalized learning. This course is designed specifically for research and higher education professionals interested in harnessing AI’s power to improve teaching and learning and advance institutional innovation.

Participants will explore the core principles behind LLMs and Retrieval Augmented Generation (RAG), gain hands-on experience building AI-driven tools, and learn how to scale these applications using cloud platforms. By the end of this course, you’ll be able to create intelligent applications that accelerate research, personalize education, and leverage real-time data—all while staying aligned with the latest AI and cloud trends.

Skill Level and Target Audience

Introductory to Intermediate
Research IT, Enterprise IT

Workshop Objectives

Learn how to use embeddings to search for meanings and context within text and explore different embedding models.
Understand how to perform text searches using vector databases that leverage embeddings for efficient information retrieval.
Discover what Large Language Models (LLMs) are and how to use them via APIs or locally deployed instances.
Gain hands-on experience building applications integrating LLMs for text generation and natural language understanding tasks.
Create Retrieval-Augmented Generation (RAG) applications that combine real-time data retrieval with LLM-driven text generation.
Learn how to containerize and deploy your AI-driven applications for scalability and efficient distribution across different environments.

About the Instructor

Sujee Maniyam is a hands-on AI practitioner, speaker, and educator who is passionate about making complex AI accessible to everyone. As a co-founder of Elephant Scale and a seasoned practitioner, Sujee has extensive experience deploying AI-driven solutions at scale, making him uniquely qualified to guide learners through building intelligent applications using large language models (LLMs).

Workshop Details

- Understanding embeddings allows models to understand relationships for tasks like text similarity, clustering, and classification.
- Embedding models to generate vectors is helpful for tasks like search and recommendation.
- Semantic text search using embeddings.
- Evaluating embedding models (e.g., BERT vs. Word2Vec, etc.) to assess accuracy, speed, and their ability to capture contextual or domain-specific meanings.
- Introduction to vector databases
- Getting started with MongoDB Atlas
- Loading data into databases and populating embeddings
- Vector search with databases and embeddings
- Introduction to LLMs
- Access LLMs via APIs
  - Utilize APIs from platforms like OpenAI or Hugging Face to easily integrate and interact with pre-trained LLMs in your applications without complex setups.
- Run LLMs locally using llama-cpp, oolama, lm-studio, jen, etc.
  - Learn how to deploy LLMs on your local machine using frameworks such as llama-cpp, oolama, lm-studio, and jen to reduce cloud reliance and increase control over your models.
- Experiment with different LLMs
  - Explore and compare open-source LLMs to understand their performance, features, and suitability for specific tasks or environments.
- Leverage specialized frameworks for seamless local LLM integration
  - By using tools like llama-cpp, llama-index, and Langchain, you can efficiently load, manage, and query Large Language Models (LLMs) on local machines.
- Deploy LLMs on your local environment for faster, private AI processing
- Test and compare different open-source LLMs to find the best fit for your project
  - Experimenting with models like Mistral, Llama, and others allows you to evaluate their performance in text generation, summarization, and question-answering tasks.
- Integrate LLM capabilities into your application by using pre-trained models like GPT to add features customized for your specific use case, such as natural language understanding, text generation, or summarization.
- Leverage frameworks like Streamlit and Flask
  - Quickly build and deploy interactive web applications with minimal coding, allowing real-time interactions with your LLM-powered features.
- Prototyping and scaling
  - Use these frameworks to create sample applications for testing and showcasing AI functionality, then scale them up for broader use in academic, research, or business environments, incorporating LLM-driven insights into everyday tasks.
- Query PDF documents using LLMs
  - Utilize LLMs to process and query the content of PDF documents, extracting relevant information and insights through natural language queries.
- Index documents with embeddings
  - Create embeddings for documents, turning their content into searchable vectors that capture the semantic meaning for efficient querying.
- Use various embedding models (OpenAI, Mistral, open source models)
  - Leverage different embedding models from providers like OpenAI, Mistral, and other open-source options to convert documents into vector representations.
- Query the documents using various LLMs (OpenAI, Mistral, LLama)
  - Employ LLMs such as OpenAI’s GPT, Mistral, or LLama to interpret and respond to user queries, retrieving relevant sections from indexed documents.
- Set up and run Retrieval-Augmented Generation (RAG) applications to combine real-time data retrieval with LLM-powered text generation for more accurate, context-aware outputs.
- Use containerization tools like Docker to package your RAG application, ensuring consistency, portability, and ease of deployment across different environments.
- Efficiently serve your LLMs or RAG models via APIs or frameworks, enabling real-time access and interactions with the models in your application.
- Implement strategies like load balancing, distributed computing, and auto-scaling to ensure your application can handle increased demand while maintaining performance.
- Design an AI-powered tool tailored to your institution’s needs, such as an intelligent research assistant or personalized learning platform.

Featured Services

NET+ Cloud Services

Network Services

InCommon Services

Developing Intelligent Systems with LLMs and RAG

Workshop Information

Workshop Overview

Workshop Details

If you have further questions, please contact us.