Fine-Tuning Domain Models: Does Better Data Beat Bigger Models? – #SFTW

​Join AWS and Sahara AI for an in-depth technical discussion on building domain-specific models with high-quality data.

​As models scale, it’s clear: bigger isn’t always better. The true challenge lies in sourcing, curating, and validating the bespoke datasets that actually improve accuracy, safety, and performance for domain LLMs.

​Our technical experts will share practical insights on:

• ​Candid lessons learned from building domain-specific models

• ​Real-world case studies from industry and academia

• ​Forward-looking strategies to overcome the data bottleneck in fine-tuning

• ​Best practices for balancing data quality, cost, and scalability in production pipelines

​Designed for technical leaders and senior engineers, this session will focus on actionable strategies and implementation approaches rather than theoretical frameworks.

​Agenda

• ​5:00-5:30 l Networking Break

• ​5:30-5:40 l Intro by AWS and Sahara AI

• ​5:40-6:15 l Demos and Case Studies

  • ​Sahara ai - Head of Product, James Constentine

• ​6:15 -7:15 l Panel - Building Domain Models: Data Struggles and Success Stories with Q+A

  • ​Hosted by Jean Malha / AWS

    • ​Anthropic - Rohan Taori

    • ​Sanas.ai - VP of ML/AI Data, Kiran Lokeswarappa

    • ​Sahara ai - Head of Product, James Constentine

    • ​Pinterest - Senior Director of Engineering, Bo Zhao

• ​7:15 -8:00 l Cocktail Reception

​**ID required for entry into the building**