12-05, 12:30–14:00 (UTC), AI/ML Track
LLMs offer powerful capabilities, but deploying them effectively in production remains a challenge for conversational AI and Chatbot applications, especially when it comes to minimizing hallucinations and ensuring accurate responses. In this 90-minute hands-on tutorial, we’ll explore building conversational AI systems using CALM and Rasa. CALM (Conversational AI Language Model) combines traditional conversational AI techniques with LLMs, separating conversational ability from business logic execution to deliver reliable, cost efficient, and scalable solutions. Unlike LLMs that handle both sides of the conversation, CALM focuses on user understanding with predefined business logic. This approach not only accelerates development but also enhances cost efficiency, scalability and reliability. By focusing on predefined business logic with CALM, you’ll gain the ability to build sophisticated, scalable systems faster. You’ll also learn how to use fine-tuned, open-weight models, such as llama 8b to power your AI assistant.
Participants will learn how to use CALM for business logic and Rasa for dialogue management, with practical insights, code examples, and best practices. Materials will be provided via a GitHub repository with a GitHub Codespace for easy access and execution.
This tutorial will guide participants through the process of building conversational AI systems using CALM and Rasa. We'll cover:
Introduction to CALM and Rasa (15 minutes):
Overview of CALM and its benefits in integrating traditional conversational AI with LLMs.
Introduction to Rasa and its role in dialogue management.
Setting Up the Environment (5 minutes):
Instructions for accessing the GitHub repository and GitHub Codespace.
Setting up the development environment.
Building a Travel Assistant Conversational Agent (35 minutes):
Hands-on implementation of a travel assistant agent.
Integrating CALM and Rasa to separate conversational ability from business logic execution.
Testing, Evaluation, and Optimization (15 minutes):
Techniques for testing the conversational agent.
Evaluating the system using end-to-end tests.
Best practices for optimizing performance and cost, including using a fine-tuned llama 8b models for your assistant.
Q&A and Wrap-Up (20 minutes):
Open floor for questions.
Summary of key takeaways and next steps.
Attendees should have a basic understanding of Python. All materials will be distributed via a GitHub repository with a GitHub Codespace, so participants can execute all tasks without local installations.
Previous knowledge expected
Dr. Alan Nichol is co-founder & CTO of Rasa, a widely-used open source platform for conversational AI, and has authored research papers, technical blog posts, and online courses on the subject of conversational AI. He is credited with popularizing the idea of building chat and voice bots that do not rely on 'intents', and is one of the creators of the Rasa framework. Alan holds a PhD in machine learning from the University of Cambridge, and in 2024 was named one of Europe's 100 most influential people in AI.
Hugo Bowne-Anderson is an independent data and AI consultant with extensive experience in the tech industry. He is the host of the industry Vanishing Gradients, where he explores cutting-edge developments in data science and artificial intelligence.
As a data scientist, educator, evangelist, content marketer, and strategist, Hugo has worked with leading companies in the field. His past roles include Head of Developer Relations at Outerbounds, a company committed to building infrastructure for machine learning applications, and positions at Coiled and DataCamp, where he focused on scaling data science and online education respectively.
Hugo's teaching experience spans from institutions like Yale University and Cold Spring Harbor Laboratory to conferences such as SciPy, PyCon, and ODSC. He has also worked with organizations like Data Carpentry to promote data literacy.
His impact on data science education is significant, having developed over 30 courses on the DataCamp platform that have reached more than 3 million learners worldwide. Hugo also created and hosted the popular weekly data industry podcast DataFramed for two years.
Committed to democratizing data skills and access to data science tools, Hugo advocates for open source software both for individuals and enterprises.
Data Scientist, developer, and educator with a passion for enabling developers to build great applications and turn data into meaningful insights and innovative products. With over the 10 years spent in Data Science and Developer Relations for AI and Web3 spaces, Justina has been focusing on empowering developers around he world to build better applications and products.