Darryl Ruggles
@darryl-ruggles.cloud
Principal Cloud Solutions Architect @ Ciena - AWS Community Builder
Serverless, Event-Driven Architecture, AWS, Kubernetes, Rust, Terraform, Security, DevOps, FinOps, MLOps, Maker
https://darryl-ruggles.cloud
https://www.linkedin.com/in/darryl-ruggles
Serverless, Event-Driven Architecture, AWS, Kubernetes, Rust, Terraform, Security, DevOps, FinOps, MLOps, Maker
https://darryl-ruggles.cloud
https://www.linkedin.com/in/darryl-ruggles
Pinned
2025 was an amazing year for me and our industry. I finally got to meet so many people including @danielleheberling.xyz, @readysetcloud.io, and @andmore.dev from the @believeinsls.bsky.social community. They handed me an award I'd received virtually just over a year ago.
trace ID approach for correlating data across services is the key to debugging.
Dippu Kumar Singh shows a layout for using these tools and storing the data in S3 for cost savings. Check it out!
(3/3)
Powered by @nexus_share
Dippu Kumar Singh shows a layout for using these tools and storing the data in S3 for cost savings. Check it out!
(3/3)
Powered by @nexus_share
January 13, 2026 at 6:53 PM
trace ID approach for correlating data across services is the key to debugging.
Dippu Kumar Singh shows a layout for using these tools and storing the data in S3 for cost savings. Check it out!
(3/3)
Powered by @nexus_share
Dippu Kumar Singh shows a layout for using these tools and storing the data in S3 for cost savings. Check it out!
(3/3)
Powered by @nexus_share
might pass through ten services before something breaks. It's genuinely hard.
Setting up an architecture for observability using tools like OpenTelemetry for metrics and traces, Fluent Bit for logs, and structured logging to tie it all together is a good approach. The (2/3)
Setting up an architecture for observability using tools like OpenTelemetry for metrics and traces, Fluent Bit for logs, and structured logging to tie it all together is a good approach. The (2/3)
January 13, 2026 at 6:53 PM
might pass through ten services before something breaks. It's genuinely hard.
Setting up an architecture for observability using tools like OpenTelemetry for metrics and traces, Fluent Bit for logs, and structured logging to tie it all together is a good approach. The (2/3)
Setting up an architecture for observability using tools like OpenTelemetry for metrics and traces, Fluent Bit for logs, and structured logging to tie it all together is a good approach. The (2/3)
https://dzone.com/articles/architecting-observability-in-kubernetes-with-otel-fluent-bit
Troubleshooting in Kubernetes can feel like chasing a needle through a haystack that keeps rebuilding itself. Pods come and go, logs vanish with them, and a single user request 🧵
Troubleshooting in Kubernetes can feel like chasing a needle through a haystack that keeps rebuilding itself. Pods come and go, logs vanish with them, and a single user request 🧵
January 13, 2026 at 6:53 PM
https://dzone.com/articles/architecting-observability-in-kubernetes-with-otel-fluent-bit
Troubleshooting in Kubernetes can feel like chasing a needle through a haystack that keeps rebuilding itself. Pods come and go, logs vanish with them, and a single user request 🧵
Troubleshooting in Kubernetes can feel like chasing a needle through a haystack that keeps rebuilding itself. Pods come and go, logs vanish with them, and a single user request 🧵
Check out this article from Brooke Jamieson on building these with AWS tools like Strands and AgentCore. (3️⃣/3️⃣)
January 13, 2026 at 2:10 AM
Check out this article from Brooke Jamieson on building these with AWS tools like Strands and AgentCore. (3️⃣/3️⃣)
Context graphs are an interesting take on this. Rather than just logging events, they capture decision traces. What was decided, why it made sense, what alternatives were considered, and how it all connected to customers and systems. (2️⃣/3️⃣)
January 13, 2026 at 2:09 AM
Context graphs are an interesting take on this. Rather than just logging events, they capture decision traces. What was decided, why it made sense, what alternatives were considered, and how it all connected to customers and systems. (2️⃣/3️⃣)
dev.to/aws/why-ai-a...
It's all about the context! We spent years building dashboards that showed what was happening but never why. Now we're building AI agents and hitting the same wall. The reasoning behind decisions, the context that actually matters, still isn't being captured anywhere.(1️⃣/3️⃣)
🧵
It's all about the context! We spent years building dashboards that showed what was happening but never why. Now we're building AI agents and hitting the same wall. The reasoning behind decisions, the context that actually matters, still isn't being captured anywhere.(1️⃣/3️⃣)
🧵
Why AI Agents Need Context Graphs (And How to Build One with AWS)
Back in 2020, I wrote that dashboards tell you what's happening, but not why. That the rush to be...
dev.to
January 13, 2026 at 2:09 AM
dev.to/aws/why-ai-a...
It's all about the context! We spent years building dashboards that showed what was happening but never why. Now we're building AI agents and hitting the same wall. The reasoning behind decisions, the context that actually matters, still isn't being captured anywhere.(1️⃣/3️⃣)
🧵
It's all about the context! We spent years building dashboards that showed what was happening but never why. Now we're building AI agents and hitting the same wall. The reasoning behind decisions, the context that actually matters, still isn't being captured anywhere.(1️⃣/3️⃣)
🧵
config values and are more flexible if you want to do something that isn't the most common.
Yuiko Koyanagi shows a number of CDK constructs incl creating VPCs, ECS Clusters, Fargate tasks, DynamoDB tables, and more. If you haven't used the CDK before then maybe you will learn something new. (3️⃣/3️⃣)
Yuiko Koyanagi shows a number of CDK constructs incl creating VPCs, ECS Clusters, Fargate tasks, DynamoDB tables, and more. If you haven't used the CDK before then maybe you will learn something new. (3️⃣/3️⃣)
January 12, 2026 at 10:20 PM
config values and are more flexible if you want to do something that isn't the most common.
Yuiko Koyanagi shows a number of CDK constructs incl creating VPCs, ECS Clusters, Fargate tasks, DynamoDB tables, and more. If you haven't used the CDK before then maybe you will learn something new. (3️⃣/3️⃣)
Yuiko Koyanagi shows a number of CDK constructs incl creating VPCs, ECS Clusters, Fargate tasks, DynamoDB tables, and more. If you haven't used the CDK before then maybe you will learn something new. (3️⃣/3️⃣)
The CDK supports abstractions of all the major AWS components at various levels of configurability/opinionation. Some of these use default values for almost everything and allow you to just write a couple of lines and have many things setup. Other constructs require you to enter most of the.. (2️⃣/3️⃣)
January 12, 2026 at 10:20 PM
The CDK supports abstractions of all the major AWS components at various levels of configurability/opinionation. Some of these use default values for almost everything and allow you to just write a couple of lines and have many things setup. Other constructs require you to enter most of the.. (2️⃣/3️⃣)
dev.to/toffy/hostin...
There are many tools to use for Infrastructure as Code (IaC). I typically use Terraform or the Serverless Application Model (SAM) but the AWS Cloud Development Kit (CDK) is widely used and offers a more integrated experience. Below is an interesting example of CDK. (1️⃣/3️⃣)
🧵
There are many tools to use for Infrastructure as Code (IaC). I typically use Terraform or the Serverless Application Model (SAM) but the AWS Cloud Development Kit (CDK) is widely used and offers a more integrated experience. Below is an interesting example of CDK. (1️⃣/3️⃣)
🧵
dev.to
January 12, 2026 at 10:19 PM
dev.to/toffy/hostin...
There are many tools to use for Infrastructure as Code (IaC). I typically use Terraform or the Serverless Application Model (SAM) but the AWS Cloud Development Kit (CDK) is widely used and offers a more integrated experience. Below is an interesting example of CDK. (1️⃣/3️⃣)
🧵
There are many tools to use for Infrastructure as Code (IaC). I typically use Terraform or the Serverless Application Model (SAM) but the AWS Cloud Development Kit (CDK) is widely used and offers a more integrated experience. Below is an interesting example of CDK. (1️⃣/3️⃣)
🧵
practical guide with clear explanations of why each component matters.
Gajanan Chandgadkar has put together a good intro. (3/3)
Powered by @nexus_share
Gajanan Chandgadkar has put together a good intro. (3/3)
Powered by @nexus_share
January 12, 2026 at 5:06 PM
practical guide with clear explanations of why each component matters.
Gajanan Chandgadkar has put together a good intro. (3/3)
Powered by @nexus_share
Gajanan Chandgadkar has put together a good intro. (3/3)
Powered by @nexus_share
What context? When did it happen? Without tracing, debugging becomes frustrating guesswork.
The example below covers setting up Langfuse locally with Docker, building a traced FastAPI chatbot, and understanding how traces, spans, and metrics work together. It's a (2/3)
The example below covers setting up Langfuse locally with Docker, building a traced FastAPI chatbot, and understanding how traces, spans, and metrics work together. It's a (2/3)
January 12, 2026 at 5:06 PM
What context? When did it happen? Without tracing, debugging becomes frustrating guesswork.
The example below covers setting up Langfuse locally with Docker, building a traced FastAPI chatbot, and understanding how traces, spans, and metrics work together. It's a (2/3)
The example below covers setting up Langfuse locally with Docker, building a traced FastAPI chatbot, and understanding how traces, spans, and metrics work together. It's a (2/3)
https://medium.com/@gajaoncloud/traces-spans-observability-making-your-llm-app-a-glass-box-d2c9597db9a1
Observability matters just as much for LLM apps as it does for traditional systems. A user reports a strange response and you're left guessing about which prompt? 🧵
Observability matters just as much for LLM apps as it does for traditional systems. A user reports a strange response and you're left guessing about which prompt? 🧵
January 12, 2026 at 5:06 PM
https://medium.com/@gajaoncloud/traces-spans-observability-making-your-llm-app-a-glass-box-d2c9597db9a1
Observability matters just as much for LLM apps as it does for traditional systems. A user reports a strange response and you're left guessing about which prompt? 🧵
Observability matters just as much for LLM apps as it does for traditional systems. A user reports a strange response and you're left guessing about which prompt? 🧵
The recent announcement of Durable Lambda Functions gives more options, but Step Functions are still a great choice for workflow orchestration, Pankaj Makhijani shows the code and includes deployment steps. (3️⃣/3️⃣)
January 12, 2026 at 4:52 AM
The recent announcement of Durable Lambda Functions gives more options, but Step Functions are still a great choice for workflow orchestration, Pankaj Makhijani shows the code and includes deployment steps. (3️⃣/3️⃣)
Below is an example of a complete order system for this including: Lambda functions for each step, parallel execution for email and analytics, automatic retries, and proper error handling. The visual workflow with Step Functions makes debugging easier. (2️⃣/3️⃣)
January 12, 2026 at 4:52 AM
Below is an example of a complete order system for this including: Lambda functions for each step, parallel execution for email and analytics, automatic retries, and proper error handling. The visual workflow with Step Functions makes debugging easier. (2️⃣/3️⃣)
medium.com/aws-in-plain...
Order processing involves a lot of moving parts including inventory checks, payments, notifications, and shipping. Coordinating all of that reliably without a mess of tangled code is harder than it sounds. One approach to handle this is AWS Step Functions. (1️⃣/3️⃣)
🧵
Order processing involves a lot of moving parts including inventory checks, payments, notifications, and shipping. Coordinating all of that reliably without a mess of tangled code is harder than it sounds. One approach to handle this is AWS Step Functions. (1️⃣/3️⃣)
🧵
AWS Step Functions: Build an Order Processing System
Complete guide with Lambda functions, error handling, and real-world examples
medium.com
January 12, 2026 at 4:52 AM
medium.com/aws-in-plain...
Order processing involves a lot of moving parts including inventory checks, payments, notifications, and shipping. Coordinating all of that reliably without a mess of tangled code is harder than it sounds. One approach to handle this is AWS Step Functions. (1️⃣/3️⃣)
🧵
Order processing involves a lot of moving parts including inventory checks, payments, notifications, and shipping. Coordinating all of that reliably without a mess of tangled code is harder than it sounds. One approach to handle this is AWS Step Functions. (1️⃣/3️⃣)
🧵
Róbert Zsótér is clear about what this is and isn't. It's for break-glass scenarios and training, not daily operations. If it becomes convenient then it's being misused. A code repo is included. Check it out and see if this approach makes sense in your environment. (3️⃣/3️⃣)
January 11, 2026 at 9:46 PM
Róbert Zsótér is clear about what this is and isn't. It's for break-glass scenarios and training, not daily operations. If it becomes convenient then it's being misused. A code repo is included. Check it out and see if this approach makes sense in your environment. (3️⃣/3️⃣)
The approach runs a web terminal inside the cluster using a dedicated ServiceAccount. All commands go through the Kubernetes API with standard RBAC and audit logging. There are no bastion hosts, no distributed credentials and no permanent exposure. (2️⃣/3️⃣)
January 11, 2026 at 9:46 PM
The approach runs a web terminal inside the cluster using a dedicated ServiceAccount. All commands go through the Kubernetes API with standard RBAC and audit logging. There are no bastion hosts, no distributed credentials and no permanent exposure. (2️⃣/3️⃣)
medium.com/@zs77.robert...
Kubernetes access often forces a choice between two bad options: always-on tooling with too much exposure, or locked-down environments where troubleshooting becomes impossible. This article explores a middle path for temporary, controlled access. (1️⃣/3️⃣)
🧵
Kubernetes access often forces a choice between two bad options: always-on tooling with too much exposure, or locked-down environments where troubleshooting becomes impossible. This article explores a middle path for temporary, controlled access. (1️⃣/3️⃣)
🧵
Browser-Based kubectl Access: Managing Kubernetes Without Bastion Hosts or Heavy Tools
Accessing a Kubernetes cluster securely has never been just a tooling problem. It is a question of control, visibility, and accountability.
medium.com
January 11, 2026 at 9:45 PM
medium.com/@zs77.robert...
Kubernetes access often forces a choice between two bad options: always-on tooling with too much exposure, or locked-down environments where troubleshooting becomes impossible. This article explores a middle path for temporary, controlled access. (1️⃣/3️⃣)
🧵
Kubernetes access often forces a choice between two bad options: always-on tooling with too much exposure, or locked-down environments where troubleshooting becomes impossible. This article explores a middle path for temporary, controlled access. (1️⃣/3️⃣)
🧵
scanning. Each catches different failure modes. It's a layered approach which you can consider. Whether it's worth doing this in your case will depend (as usual). This includes some things I hadn't fully thought through on chart testing. (3/3)
Powered by @nexus_share
Powered by @nexus_share
January 11, 2026 at 6:35 PM
scanning. Each catches different failure modes. It's a layered approach which you can consider. Whether it's worth doing this in your case will depend (as usual). This includes some things I hadn't fully thought through on chart testing. (3/3)
Powered by @nexus_share
Powered by @nexus_share
test charts though. Below is a case of a real outage that took four hours to resolve because RBAC changes weren't caught before production.
Alexandre Vazquez shows an approach for testing including syntax, schema validation, unit tests, integration tests, and security (2/3)
Alexandre Vazquez shows an approach for testing including syntax, schema validation, unit tests, integration tests, and security (2/3)
January 11, 2026 at 6:35 PM
test charts though. Below is a case of a real outage that took four hours to resolve because RBAC changes weren't caught before production.
Alexandre Vazquez shows an approach for testing including syntax, schema validation, unit tests, integration tests, and security (2/3)
Alexandre Vazquez shows an approach for testing including syntax, schema validation, unit tests, integration tests, and security (2/3)
https://dev.to/alexandrev/why-helm-chart-testing-matters-and-how-to-choose-your-tools-4mhk
Helm is a key tool for building Kubernetes apps. Most teams test application code rigorously but treat Helm charts as "just configuration." There are approaches you can use to 🧵
Helm is a key tool for building Kubernetes apps. Most teams test application code rigorously but treat Helm charts as "just configuration." There are approaches you can use to 🧵
January 11, 2026 at 6:35 PM
https://dev.to/alexandrev/why-helm-chart-testing-matters-and-how-to-choose-your-tools-4mhk
Helm is a key tool for building Kubernetes apps. Most teams test application code rigorously but treat Helm charts as "just configuration." There are approaches you can use to 🧵
Helm is a key tool for building Kubernetes apps. Most teams test application code rigorously but treat Helm charts as "just configuration." There are approaches you can use to 🧵
Check out the article from Mahmoud Yasien. (3️⃣/3️⃣)
January 11, 2026 at 2:33 AM
Check out the article from Mahmoud Yasien. (3️⃣/3️⃣)
GitHub Actions to get this up and running.
The example below shows the setup of EKS w/Terraform, uses the SSM CSI driver to inject configuration into pods, packages the app with Helm, and sets up a GitOps approach with Argo CD. GitHub Actions handles the CI side with automated image builds. (2️⃣/3️⃣)
The example below shows the setup of EKS w/Terraform, uses the SSM CSI driver to inject configuration into pods, packages the app with Helm, and sets up a GitOps approach with Argo CD. GitHub Actions handles the CI side with automated image builds. (2️⃣/3️⃣)
January 11, 2026 at 2:33 AM
GitHub Actions to get this up and running.
The example below shows the setup of EKS w/Terraform, uses the SSM CSI driver to inject configuration into pods, packages the app with Helm, and sets up a GitOps approach with Argo CD. GitHub Actions handles the CI side with automated image builds. (2️⃣/3️⃣)
The example below shows the setup of EKS w/Terraform, uses the SSM CSI driver to inject configuration into pods, packages the app with Helm, and sets up a GitOps approach with Argo CD. GitHub Actions handles the CI side with automated image builds. (2️⃣/3️⃣)
medium.com/@mhmudlll4xm...
Seeing examples always helps with learning. #FastAPI is a nice way to host your API. Setting it up in Kubernetes with secrets management and automated CI/CD involves a few moving parts. On AWS you can use EKS along with Terraform, Helm, SSM, #ArgoCD, and.. (1️⃣/3️⃣)
🧵
Seeing examples always helps with learning. #FastAPI is a nice way to host your API. Setting it up in Kubernetes with secrets management and automated CI/CD involves a few moving parts. On AWS you can use EKS along with Terraform, Helm, SSM, #ArgoCD, and.. (1️⃣/3️⃣)
🧵
Deploy FastAPI on AWS EKS with Terraform, SSM, and GitOps — A Step-by-Step
In this Write-up, you’ll learn how to build a secure, observable, and fully automated cloud-native platform on AWS using.
medium.com
January 11, 2026 at 2:32 AM
medium.com/@mhmudlll4xm...
Seeing examples always helps with learning. #FastAPI is a nice way to host your API. Setting it up in Kubernetes with secrets management and automated CI/CD involves a few moving parts. On AWS you can use EKS along with Terraform, Helm, SSM, #ArgoCD, and.. (1️⃣/3️⃣)
🧵
Seeing examples always helps with learning. #FastAPI is a nice way to host your API. Setting it up in Kubernetes with secrets management and automated CI/CD involves a few moving parts. On AWS you can use EKS along with Terraform, Helm, SSM, #ArgoCD, and.. (1️⃣/3️⃣)
🧵
data flow and process the data without having to provision any servers yourself.
This example below from Sanjay shows ingestion of data from a stock exchange and processing of it using various AWS serverless components. (2/2)
Powered by @nexus_share
This example below from Sanjay shows ingestion of data from a stock exchange and processing of it using various AWS serverless components. (2/2)
Powered by @nexus_share
January 10, 2026 at 4:52 PM
data flow and process the data without having to provision any servers yourself.
This example below from Sanjay shows ingestion of data from a stock exchange and processing of it using various AWS serverless components. (2/2)
Powered by @nexus_share
This example below from Sanjay shows ingestion of data from a stock exchange and processing of it using various AWS serverless components. (2/2)
Powered by @nexus_share