HALL 2 - Tech Talks / Workshops

Exploring the alchemy of transforming raw data into compelling narratives, this talk delves into the art and science of crafting captivating stories from the vast landscape of information.

Delve into the potential chaos lurking in unmonitored models and pipelines as we explore unforeseen vulnerabilities, data drift, and the silent erosion of model performance. Learn actionable strategies to preemptively safeguard against pitfalls, ensuring the robustness and reliability of your AI systems. Don’t let the unseen jeopardize your success—join us for a vital discussion on proactive monitoring and mitigation.

This talk will cover the practical lessons learned from the speaker’s experience in developing large-scale ML systems for search and recommendations. Both search and recommendations are the most impactful and yet the most complex applications of ML in the industry. Interestingly, most of these challenges do not arise from the algorithmic complexity but from the constraints related to business, technology and people. The talk will focus on a few key challenges and potential ways to overcome or avoid those.

The talk will cover the below aspects:

a) Semantic Search Applications at a Glance b) Application Focus: Semantic Search for SEO c) Application Focus: Semantic Search for UX d) RAG Applications at a Glance e) Application Focus: RAG for Answering Questions f) Application Focus: RAG for Verifying Statements g) RAG as a Remedy for Hallucinations and Outdated Models h) Evaluating Semantic Search and RAG i) Q&A Session

The talk will focus on the below aspects:

1. The importance and relevance of Large Language Models (LLMs) in today’s artificial intelligence landscape.

2. Optimizing the performance of LLMs: Techniques such as prompt engineering, fine-tuning, and Retrieval-Augmented Generation (RAG) will be discussed.

3. Challenges and ethical concerns associated with using LLMs in real-world applications. will address the mitigations and methods to counter these issues.

4. Tackling the issue of trust in language models, with a focus on developing models that are reliable and less prone to hallucinations. We will explore strategies and techniques to mitigate hallucinations in language models.

5. The impact of LLMs on enhancing productivity, customer satisfaction, and decision-making processes, along with discussing the limitations and potential areas for future improvement in the application of LLMs across various domains.