Unsupervised Learning
Ep 41: Head of AI at Snowflake Baris Gultekin on Why They Built Their Own LLM, Governance as a Moat, and the Most Common Enterprise Use-Cases
Aug 28, 2024
Baris Gultekin, the Head of AI at Snowflake, oversees the development of innovative AI tools, including their proprietary LLM, Arctic. He shares insights on how Snowflake enhances business intelligence with efficient data applications and discusses the intricacies of building Arctic, including challenges and solutions. Gultekin emphasizes the significance of data governance and security in AI deployment. The conversation also touches on the dynamic between open-source and closed-source AI models, highlighting the critical role of partnerships in advancing AI capabilities.
46:51
Episode guests
AI Summary
Highlights
AI Chapters
Episode notes
Podcast summary created with Snipd AI
Quick takeaways
- Snowflake's Arctic LLM is designed for enterprise applications, enabling efficient natural language to SQL translations for BI tasks.
- Emphasizing data governance, Snowflake ensures secure AI deployments with robust frameworks and granular access controls for sensitive information.
Deep dives
Development of Arctic LLM
Snowflake developed Arctic, a large language model (LLM) specifically designed to cater to enterprise needs such as business intelligence (BI) and SQL generation. The team behind Arctic was relatively small yet comprised of talented researchers who previously worked on notable projects. They focused on building a model that could effectively translate natural language queries into SQL, which is often complicated in real-world applications. With a three to four month development timeline, Arctic was created with efficiencies that allowed it to be developed at one eighth the cost compared to similar models in the market.
