Description:

As climate change becomes an increasingly urgent issue, finding effective ways to implement sustainable practices is essential. This talk will explore how Large Language Models (LLMs) can be used to educate and provide personalized recommendations and actionable insights tailored to various roles, from data scientists and product managers to educators and policymakers. Attendees will also learn how to start with prompt engineering and iteratively enhance the application to build a robust LLM application.

We'll start by explaining the fundamentals of climate change and why sustainability is crucial. Then, we’ll demonstrate practical use cases where LLMs offer sustainability advice, followed by a deep dive into how the initial version of this application was developed just using prompt engineering. Then we’ll discuss the enhancements made with Langchain, LangGraph, and agents to reduce hallucinations and improve recommendations. Finally, we’ll cover the deployment, monitoring, and evaluation of the model in production, and provide practical tips for deploying LLM applications.

Key Takeaways:

• Understand the importance of developing sustainable systems to combat climate change.
• Understand the iterative process of building LLM Applications.
• Discover how LLMs can deliver tailored sustainability recommendations to various professional roles and industries.

Who Will Benefit:

• Professionals across different sectors looking to integrate sustainable practices into their workflows.
• Individuals interested in deploying custom LLM solutions in production environments.
• Technology enthusiasts interested in the application of AI for environmental sustainability.
• Decision-makers aiming to incorporate AI-driven strategies for sustainability in their organizations.

Talk Outline:

• Introduction to Climate Change and Sustainability (5 mins)
• Demo and Use Cases (5 mins)
• Building the First Version with Prompt Engineering (5 mins)
• Enhancing with Langchain, LangGraph, and Agents (5 mins)
• Deployment, Monitoring, and Evaluation in Production (5 mins)
• Q&A (5 mins)

Prerequisites:

• An interest in applying AI to solve real-world environmental issues.
• A basic understanding of AI and LLMs (detailed technical knowledge is not required).

Speaker Info:

I am Dipakkumar Tandel, a Data Scientist practitioner with over 5 years of experience in building large-scale, event-driven, microservice-based machine learning solutions. My expertise includes ML engineering, MLOps, deep learning, computer vision, and natural language processing. At Blue Yonder, I create scalable, production-grade model training and inference pipelines using TensorFlow, Kubeflow, and TFX. I also focus on optimizing ML infrastructure for performance and cost efficiency and deploy robust monitoring systems to ensure model reliability.

Speaker Links:

LinkedIn : https://www.linkedin.com/in/dipak-tandel/

Talk at ChennaiPy: ML Pipeline using TFX