DataHour: Building Robust RAG Applications for the Real-World

About the Event

Retrieval-Augmented-Generation (RAG) systems offer exciting possibilities for using the latest language technology in various real-life situations. Yet, bringing RAG into the real world comes with its own set of fresh challenges. This presentation takes a direct approach to solving these challenges by suggesting ways to create strong and reliable RAG systems suitable for real-world use.

We'll start by explaining what RAG is, discussing its main components, and showing how practical it is in terms of cost-effectiveness and ease of use with language models. Then, we'll look into common design mechanisms for RAG systems and discuss their working. Additionally, we'll explore recent research developments in RAG technology aimed at making it more dependable and efficient. Finally, we'll talk about common issues with RAG and propose solutions to mitigate them. Finally, we will end the discussion with tips and tricks on thoroughly evaluating complex RAG systems.

About the Speaker

Aishwarya Naresh

GenAI Lead at AWS

Aishwarya works as a GenAI tech lead at the AWS GenAI Innovation Center, leading partnerships with external AWS customer organizations, usually mid to large-sized companies. These partnerships aim to identify, develop, and evaluate cutting-edge solutions using generative AI technologies for critical business functions. With more than eight years of experience in machine learning and contributions to over 25 papers in prestigious conferences such as AAAI, ACL, CVPR, NeurIPS, and EACL, her work spans a range of topics. These include Large Scale Graph Neural Networks, Machine Translation, Multimodal Summarization, and the exploration of Social Media and Networks, Human-Centric ML, Artificial Social Intelligence, and Code-Mixing. In addition to her professional work, Aishwarya has been involved in mentoring several Master's and PhD students in the aforementioned areas. Aishwarya has also designed multiple online courses covering Generative AI concepts, which have drawn in over 3,000 students. Additionally, she frequently shares simplified insights on complex generative AI research on LinkedIn, making it easier for more people to understand and use.

Participate in discussion

Registration Details

5353

Registered

DataHour: Building Robust RAG Applications for the Real-World