Google

Run Ray on TPU, Part 1: The foundations

Enabling distributed Python workloads to run efficiently on Google Cloud TPUs while respecting the strict networking requirements of multi-host TPU slices that must maintain connectivity over their Inter-Chip Interconnect (ICI).

distributed-systems ml-systems
5 min
Google

Run Ray on TPU, Part 2: Ray AI libraries

Efficiently running distributed AI workloads across Google's TPU slices while managing data loading bottlenecks and cross-slice coordination complexity.

distributed-systems ml-systems
5 min
Google

Scaling Agentic RL: High-Throughput Agentic Training with Tunix

TPU idling bottlenecks during multi-turn, tool-using LLM agent training caused by network I/O and environment step latencies reducing hardware throughput.

ml-systems distributed-systems
5 min
Google

Systems Engineering Playbook: Optimizing Qwen 3.5-397B MoE on Ironwood (TPU7x)

How to efficiently serve a 397-billion parameter Mixture-of-Experts model on Ironwood TPU7x hardware while overcoming severe sharding constraints and cross-device communication bottlenecks.

distributed-systems ml-systems
5 min
Google

We terminated a TPU mid-training and it recovered in seconds: Introduction to elastic training with MaxText

Distributed AI training jobs fail completely when a single machine fails, requiring expensive full-workload restarts from scratch.

distributed-systems ml-systems
5 min
Google

Build agentic full-stack apps with Genkit

Developers were struggling with the complex implementation details of building conversational AI applications, including managing message history, tool invocation loops, and real-time streaming across full-stack architectures.

api-design ml-systems
5 min
Google

Build reliable multi-agent applications with ADK Go 2.0. Discover our new graph-based workflow engine, built-in human-in-the-loop, and dynamic orchestration

Developers needed a unified runtime to compose and orchestrate complex multi-agent applications without managing separate execution models for single-agent versus multi-agent workflows.

microservices distributed-systems
5 min
Google

Build Cross-Language Multi-Agent Team with Google’s Agent Development Kit and A2A

Enabling multiple AI agents written in different programming languages to collaborate seamlessly on complex tasks like contract compliance verification.

api-design microservices
3 min
Google

Announcing the Agentic Resource Discovery specification

Agents need a standardized way to discover, identify, and verify available tools and skills across distributed systems without centralized coordination.

api-design distributed-systems
3 min
Google

How A2A is Building a World of Collaborative Agents

AI agents need a secure, flexible way to collaborate and hand off tasks without the constraints and context pollution of traditional API-based communication.

api-design distributed-systems
5 min
Google

Unlocking the Power of the TPU Stack: Introducing our new Developer Hub

Developers lack centralized, practical resources to optimize machine learning workloads and fully utilize the performance capabilities of Google Cloud TPUs.

ml-systems distributed-systems
5 min
Google

How the community trained Gemma to "Think" with Tunix and TPUs

How to enable developers with limited compute budgets to transform small base language models into capable reasoning engines through efficient training techniques.

ml-systems distributed-systems
5 min
Google

An important update: Transitioning Gemini CLI to Antigravity CLI

Google needed to unify fragmented AI terminal tooling by consolidating the community-focused Gemini CLI into a more scalable, agent-first platform capable of handling complex multi-agent workflows.

api-design microservices
5 min
Google

Build Long-running AI agents that pause, resume, and never lose context with ADK

Building production-grade AI agents that can maintain context and state across long-running enterprise workflows spanning days or weeks without losing information during idle periods or server restarts.

api-design distributed-systems
5 min
Google

Empowering Service Providers and Hardware Partners with Gemini for Home

How can Google enable third-party service providers and hardware manufacturers to build intelligent smart home experiences without requiring deep AI/ML expertise or significant R&D investment?

api-design ml-systems
5 min
Google

MaxText Expands Post-Training Capabilities: Introducing SFT and RL on Single-Host TPUs

Enabling efficient post-training of large language models on single-host TPU configurations without requiring complex multi-host distributed setups.

ml-systems distributed-systems
5 min
Google

One Year of Innovation: Celebrating 100k Members in the Google Cloud x NVIDIA Developer Community

Developers needed accessible infrastructure, resources, and structured learning pathways to effectively build and optimize AI applications using GPUs and large language models at scale.

api-design ml-systems
5 min
Google

Speeding Up AI: Bringing Google Colossus to PyTorch via GCSFS and Rapid Bucket

AI training pipelines were bottlenecked by slow data I/O when accessing training datasets stored in Google Cloud, limiting throughput and increasing total training time.

storage-systems ml-systems
5 min