Google By The Bay

Google’s relationship with the By The Bay community runs deep. Over the years, their engineers and researchers have brought some of the most practical talks to our stage—ranging from machine learning security to large-scale geospatial analysis and cloud infrastructure.

RSVP

As we look ahead to AI By The Bay 2025, we’re once again joined by an exceptional speaker from Google: Mihai Maruseac.

This year, Mihai will speak on one of the most urgent challenges in the machine learning ecosystem: supply chain security, exploring how Google built model-signing infrastructure using Sigstore and deployed it across Kaggle. The result is a scalable solution for verifying models and preventing tampering—something the entire ML community will benefit from.

In 2023, Mihai delivered a talk titled “Model Transparency for AI/ML Security.” He addressed the growing risks at the intersection of rapid AI model releases and a 700% surge in supply chain attacks, highlighting real-world cases of compromised models. Mihai presented solutions inspired by traditional software practices—artifact signing, provenance generation, and software bills of materials—and introduced a model transparency toolkit, a key pillar of Google’s Secure AI Framework (SAIF), aimed at securing ML models across

frameworks and hubs.

Just a few years earlier, at Scale By The Bay 2019, Michael Entin delivered a talk that helped bridge the gap between traditional data engineering and geospatial analysis. His session was aimed at SQL engineers unfamiliar with GIS, offering a hands-on introduction to using Google BigQuery GIS. He showed how anyone comfortable with SQL could start asking location-based questions and unlock insights from spatial data without needing to become a GIS expert.

Going further back, at Data By The Bay 2016, Google introduced us to the foundation of what would become BigQuery. The team walked us through the evolution of Dremel—an internal system developed to simplify large-scale data analysis—which eventually became BigQuery, one of the first fully managed cloud data warehouses. The talk showcased how Google separated compute and storage, built for speed and scale, and let engineers focus on insights instead of infrastructure. What was once an internal tool is now a cornerstone of modern data engineering.

 What’s Google Focused on Now?

Google is doubling down on its AI-first mission, embedding advanced models like Gemini 2.x across products—from Search to Workspace. Search is evolving into a conversational assistant powered by real-time, multi-threaded query analysis, while projects like Astra aim to deliver multimodal AI across devices, including AR glasses.

At the infrastructure level, Google is scaling TPUs (Trillium), exploring quantum computing breakthroughs, and expanding AI in health, education, and sustainability. With 2025 shaping up as a defining year, Google’s focus is clear: build AI tools that are useful, grounded, and trustworthy.

RSVP