The NVIDIA Summit 2024 in Mumbai brought together visionaries, innovators, and AI enthusiasts to explore the transformative potential of artificial intelligence. A few of our pioneers had the privilege of attending this premier event, diving deep into the latest advancements in AI, including generative AI, large language models (LLMs), robotics, and AI infrastructure powered by NVIDIA’s cutting-edge GPUs. Here are our insights, key learnings, and plans to incorporate these groundbreaking technologies into our work.

What were the main topics or themes discussed?

A wide range of topics and themes focusing on the latest advancements and applications of AI were discussed from which the main topics and themes revolved around, Generative AI and Large Language Models (LLMs), Robotics and Automation, AI Business Strategy, AI Infrastructure with NVIDIA’s advanced GPUs and AI workloads.

What were the most important insights or information gained from the summit?

• Graph RAG integrates knowledge graphs with retrieval-augmented generation for efficient information extraction.
• GPU acceleration enhances the performance of all components, significantly improving accuracy and speed.
• Fraud Detection: Advanced models using graph technology and GPU acceleration enhance real-time fraud and mule detection.
• Investment Scams: Customers are often tricked into making payments through enticing schemes, highlighting the need for better awareness and monitoring.
• Trustworthy AI: NVIDIA’s NeMo Guardrails enhance AI safety, security, and ethical considerations.
• Guardrails: Programmable guardrails ensure AI models stay on topic, secure, and ethical.
• Content Safety: Enhanced moderation tools like Llama Guard and Aegis improve content safety.

How was the overall experience at the summit?

The NVIDIA Summit was an enlightening experience, showcasing cutting-edge advancements in AI technology. Sessions highlighted the importance of AI safety and ethical considerations with NeMo Guardrails. The summit also emphasized the role of programmable guardrails and enhanced content safety tools such as Llama Guard and Aegis. Advanced security measures to protect AI models from threats were discussed in detail. The cherry on the cake was attending the fireside chat which provided valuable insights into the practical applications of these technologies, making the summit a comprehensive exploration of AI’s future.

How do you plan to implement these learnings in your work?

We plan to implement the learnings to enhance AI safety, improve model accuracy and strengthen content safety and moderation.

What new skills or knowledge did you acquire?

• Real-Time Payment Transaction Analysis and Fraud Detection at Population Scale by Prashant Devadiga (Head- Data Science & AI at NPCI), where we got the insights into staying ahead of emerging fraud threats and trends. The methods to scale fraud detection capabilities to manage growing transaction volumes and complexity. And enhancing fraud detection systems by leveraging cutting-edge technologies like machine learning, artificial intelligence, and data analytics
• LLM Safety and Security by Mohit Sewak Ph.D (Head of AI Researcher and Developer Relations, South Asia at NVIDIA), here we learnt how LLMs can be tricked into executing malicious calls or veering into unwanted domains and the importance of LLM Safety and security using the techniques to moderate and ensure content safety