Using Antigravity Like a Pro: CLI, SDK, and Science Skills
August 4, 2026
Online
Embedded Files
About the Session
Building on our previous exploration of the Antigravity IDE, this session dives into HPC workflow management and automation using the Antigravity CLI, SDK, and specialized Science Skills. We will demonstrate how to transition from manual scripting to autonomous multi-agent orchestration, enabling researchers to manage complex workflows across local, cluster, and cloud infrastructures with unprecedented efficiency across multiple domains.
What we'll cover:
Natural-Language Shell Pipelines: Translate simple English prompts into intricate Linux command-line operations—seamlessly combining multi-stage pipes, redirections, and process substitutions.
Automated Scientific Data & File Manipulation: Perform complex, multi-file content transformations, regex structural refactoring, and data format conversions directly across large scientific codebases and input decks.
Legacy Code Modernization: Using autonomous agents can analyze legacy Fortran or C libraries to refactor and optimize code for modern Python, CUDA, or JAX environments. This process includes rigorous mathematical verification against the original source, delivering both the updated code and a comprehensive verification artifact.
Autonomous MPI/GPU Debugging: Operating within secure, sandboxed terminals, agents can proactively identify multi-node race conditions and segmentation faults. By diagnosing root causes like mismatched MPI communications, agents can autonomously patch, recompile, and validate code, significantly reducing downtime.
Natural-Language Cluster Provisioning & Administration: Enables researchers and admins to define complex infrastructure in plain English—automatically translating requests into Terraform or Cluster Toolkit blueprints, executing Slurm health benchmarks, orchestrating self-healing multi-node MPI environments, and containerizing legacy scientific software for production-ready cloud deployment.
If you spend more time wrestling with fragile Slurm scripts, compiler flags, and cluster configs than advancing your actual research, this webinar will transform how you work. Join us to see live, multi-agent workflows in action and learn how to eliminate the daily operational friction of scientific computing—so you can spend less time fighting infrastructure and more time driving breakthroughs.
Speaker
Adnan Khaleel
Outbound Product Manager, Google Cloud
Adnan Khaleel is driven by a single goal: making powerful technology accessible to a broader range of users. He works closely with engineering teams to translate customer needs into product strategy, ensuring that tools are intuitive and easy to consume. This passion has been a constant throughout his more than 20-year career at innovators like Intel, Dell, and Cray. He has always focused on creating new growth engines—from leading Dell’s HPC democratization strategy to pioneering Cray's entry into the AI and data analytics space.
Today at Google, his core mission is to scale these solutions globally by creating effective go-to-market strategies. While fostering a vibrant partner ecosystem is one important way he achieves this, his main focus is on empowering users directly. He is always happy to connect with anyone passionate about solving big problems in this space.
Google Sites
Report abuse