Articles

Wire HTTP routes to Lambda functions using OpenAPI, Parameter Store, and Terraform — no hardcoded ARNs, no manual setup.

I’ve published a full working demo of a modern analytics stack: dbt transformations modeled with raw → staging → marts BigQuery as the cloud data warehouse Tableau dashboard visualizing business metrics Fully modular, automated, and environment-agnostic setup Explore the project: Analytics Repo on GitHub View Dashboard on Tableau Public Everything is built to be portable, extensible, and reproducible — designed both for demonstration and as a real starting point for more complex pipelines. ...

A high-level view of how Adage handles configuration-driven deployments Terraform wasn’t designed for object-oriented programming. It’s declarative, static, and prefers duplication over abstraction. But infrastructure doesn’t have to be. With the right separation of concerns, you can make your infrastructure modular, composable, and runtime-aware. You can design a system where components behave more like objects: isolated, testable, and able to change behavior based on identity and configuration. ...

Karma is an experimental system that treats infrastructure not just as state — but as consequence. What Is Karma? Karma is a new open source project that builds on the ideas behind Adage, extending them into a fully introspectable, graph-based runtime. At its core, Karma models your infrastructure as a graph of components, each defined by versioned configuration (in Git and AWS Parameter Store) and connected by runtime dependencies. Every component is: ...

The fastest way to go live with infrastructure you actually understand What if you could deploy a secure, fast, fully serverless website on AWS using just a JSON config file — and know exactly how it works under the hood? That’s the idea behind Adage: A Configuration-Driven AWS Deployment Framework that separates your infrastructure, your config, and your application code into distinct, composable layers. In this guide, you’ll see how to go from zero to live in a few minutes using the serverless-site component — and how this approach gives you clarity, repeatability, and control without locking you into someone else’s boilerplate. ...

Open source Adage: Configuration-Driven AWS Deployment Framework wasn’t built as a machine learning platform — but it enables one. Because each deployed component exposes its runtime state, dependencies, and configuration via AWS Parameter Store, the system becomes: Observable Configurable at runtime Adaptable without redeployment What Is Adaptive Runtime Behavior? In this context, it means infrastructure that responds to runtime conditions: Services that select behavior based on observed state Configuration that adjusts over time Decision logic that operates independently from deployment processes Why This Framework Supports It 1. Decoupled Runtime State Every component stores its runtime info under a consistent path: ...

This article outlines the core design principles behind a configuration-driven approach to AWS infrastructure. These principles are designed to support modularity, auditability, scalability, and low-friction deployment across environments. Together, they form a consistent model that separates infrastructure code, configuration, and runtime behavior—while keeping everything connected through predictable patterns. If you’re looking to implement infrastructure that supports dynamic environments, feature isolation, and clean service boundaries, these principles provide a framework for building systems that are both flexible and easy to manage. ...

Traditional AWS Deployments Are Holding You Back If you’ve worked with AWS long enough, you’ve likely experienced these pain points: Manually updating Terraform every time a new environment or feature is needed Hardcoded infrastructure dependencies that make it difficult to scale Multiple teams needing access to Terraform, even if they only manage configurations Lack of auditability — who changed what, and when? The traditional way of managing AWS infrastructure forces DevOps teams into bottlenecks and makes scaling complex architectures painful. ...