Building Agents at Cloud Scale
A practical path for taking MCP-style agents to AWS scale—what changes at each stage, the patterns that keep agents reliable, and demos along the way.
Hi! I'm
Antje Barth
Member of Technical Staff · Keynote Speaker · Author
As a Member of Technical Staff at Amazon AGI Labs, I drive how developers discover, learn, and build with Amazon's next-generation AI. I work cross-functionally with Research, Product, and Engineering to translate emerging model capabilities into practical frameworks and guidance that help teams ship real applications. I'm also co-author of the O'Reilly books Generative AI on AWS and Data Science on AWS.
Keynotes & Talks
A few recent talks where I share what I've learned building and scaling AI systems.
Scaling MCP
How to scale the Model Context Protocol in the cloud—serverless HTTP handlers on AWS Lambda, pragmatic context management, and where agentic services are heading next.
From Software Developer to AI Engineer
How generative AI reshapes software development—new workflows, team habits, and AWS tools that help you move from idea to production faster.
Books
O'Reilly books I co-authored on generative AI and data science.
O'Reilly
Generative AI on AWS
Build multimodal applications with foundation models, LLMs, and diffusion models. Learn fine-tuning techniques including RLHF, PEFT, and prompt engineering on Amazon Bedrock and SageMaker.
O'Reilly
Data Science on AWS
Build end-to-end ML pipelines with SageMaker, TensorFlow, and BERT. Master data engineering, model training, hyperparameter tuning, and production deployment at scale.
Online Course
A course I built and co-instructed with Dr. Andrew Ng and DeepLearning.AI.
Generative AI with Large Language Models
On-demand DeepLearning.AI course that I managed development for and co-instructed in collaboration with Dr. Andrew Ng and the team at DeepLearning.AI. We cover the full GenAI lifecycle (data and model selection through evaluation and deployment), explain transformers and fine-tuning, use scaling laws to guide trade-offs, and apply current best practices for training, tuning, inference, tooling, and deployment.
PR/Media
Podcasts, interviews, and articles where I get to talk shop.
How AWS is Helping Developers with AI Reality
What it takes to move agentic AI from hype to day-to-day developer reality—tools, safety, and a shipping mindset.
How open source and agentic AI are converging
Live at Open Source Summit Denver, a look at where open source meets agentic AI—and what that means for teams building on AWS.
Three AI-Assisted Development Skills You Can Use Today
Three practical skills you can adopt now to level up AI-assisted development without derailing your existing workflow.
The skills developers need in an AI era
A candid chat about the skills that matter as AI becomes table stakes—the rise of AI engineering and even a bit on "vibe coding."
Generative AI/ML for developers
A conversation about how developer advocates help builders navigate new GenAI workflows and bring ideas to production.
Blog
Some of my recent posts on the AWS News Blog.
Amazon Bedrock Guardrails now supports multimodal toxicity detection with image support (preview)
Launch post for image-aware content filtering in Guardrails—how to configure thresholds, and test the feature in the console and SDKs.
Introducing the next generation of Amazon SageMaker: The center for all your data, analytics, and AI
A tour of the new unified SageMaker—Unified Studio, Amazon Bedrock IDE integration, and Amazon Q woven into everyday workflows.
Introducing multi-agent collaboration capability for Amazon Bedrock (preview)
How Bedrock's supervisor-and-subagents model helps specialized agents work together on complex tasks—plus a step-by-step demo to get started.