Hunter Data Analytics

Talk Description: TBA...
While the initial capabilities of Large Language Models often center on the prompt, the real challenge lies in the architecture that surrounds them. This session explores the transition from basic RAG implementations to sophisticated, agentic workflows, focusing on the engineering required to turn static data into actionable systems.

Key discussion points:
1. Architectural Foundations: An examination of hybrid search approaches, specifically balancing semantic and lexical search to improve retrieval accuracy.

2. Agentic Workflows: A technical breakdown of the shift from static RAG to agentic systems, including patterns for multi-step reasoning and autonomous task execution.

3. The Evaluation Layer: A focus on moving beyond intuition toward systematic measurement. We will cover the implementation of key metrics, including Mean Reciprocal Rank (MRR) and Hit Rate, and how they inform continuous model improvement.

The session provides a technical roadmap for navigating the trade-offs, engineering patterns, and rigorous testing required to deploy intelligent, production-grade retrieval systems.

Speaker Bio: Nakul Bajaj has taught at Deakin University, The University of Queensland and RMIT, helping students understand the foundations of data and analytics. He is the co author of Data Centric Machine Learning with Python, which focuses on leading from data, and bridging the gap between data science and engineering.