Developing Scientific & Technical Applications with Agentic AI Using Google Antigravity

About the Session

For scientific and technical developers, the bottleneck to discovery is rarely a lack of ideas—it is the friction of writing, debugging, compiling, and optimizing complex code. The arrival of Google Antigravity 2.0, leveraging the new Gemini 3.5 Flash model optimized for coding and agentic CI, completely changes this dynamic by introducing an agent-first IDE designed specifically for the rigorous demands of advanced computing.

Join the Google Cloud Advanced Computing Community for an interactive, demo-driven session showcasing how Antigravity 2.0 transforms the software development lifecycle for high-performance computing. We will move past the basics to look at how this platform addresses the unique challenges of building highly parallelized and accelerated software.

What We Will Cover:

• The Antigravity 2.0 IDE: A deep dive into the completely redesigned environment where autonomous coding agents work natively alongside your workspace, acting as tireless, expert co-developers.

• The CLI & SDK in Action: See how to interact with the environment programmatically. We will demonstrate how to use the Antigravity CLI to trigger automated code-generation loops directly from your terminal and how the SDK lets you build custom software automation routines.

• Deploying Specialized "Science Skills": Watch how the environment utilizes domain-specific models to understand scientific equations, physics constraints, and mathematical logic—translating complex theory into syntactically correct code.

• Taming MPI and GPU Complexity: The core of our presentation will focus on real-world advanced computing code. We will demonstrate how Antigravity 2.0 accelerates the development and debugging of MPI parallel applications and GPU-accelerated applications. Watch autonomous agents analyze race conditions, locate memory leaks across multi-node execution blocks, and suggest optimal CUDA or OpenMP directives to maximize performance.

Whether you are writing distributed-memory physics simulations or tuning a massive deep-learning framework, this session will show you how to leverage Antigravity 2.0 to eliminate boilerplate, crush debugging cycles, and bring your technical software to production faster.

Speaker