November 2023 Summaries

CircleCI jobs can securely access Azure resources using OpenID Connect (OIDC) identity tokens, eliminating the need to store static credentials. The process involves creating an Azure Active Directory application and service principal, assigning roles, and adding federation credentials to make the application available to OIDC. A CircleCI context is created to store Azure-related IDs securely, which are then used in a CircleCI workflow configured in a YAML file. This setup includes using a Docker image with Azure CLI pre-installed and executing Azure login and VM listing commands with the reader role. The workflow demonstrates successful authentication and access to Azure resources using OIDC, showcasing CircleCI's capabilities for secure multi-cloud integration and deployment.

The tutorial provides a comprehensive guide on using the Amazon SageMaker Orb in conjunction with CircleCI to manage model deployment across different environments. It walks readers through setting up a Deployment Integration token and configuring Amazon SageMaker access using OIDC. The guide also explains creating an IAM role with specific permissions, setting environment variables, and configuring the CircleCI config.yml file to facilitate model deployment. Readers are instructed on using an example project repository to manage model package versions and deploy models to various environments, with steps for creating and deploying model endpoints. The tutorial emphasizes monitoring deployment progress via the CircleCI Deployments page and highlights an upcoming feature for one-click model rollbacks. Additionally, readers are directed to support resources for further assistance.

The article provides a comprehensive guide on automating deployments of a .NET application to Azure Kubernetes Service (AKS) using Microsoft Azure and CircleCI. It covers the entire process from setting up the development environment and creating a service principal, to building and pushing Docker images to Azure Container Registry (ACR), and deploying them to an AKS cluster. The article emphasizes the importance of a continuous integration/continuous delivery (CI/CD) pipeline in ensuring efficient and reliable application updates, highlighting the use of Kubernetes manifests for deployment configuration and CircleCI for managing the deployment workflow. Additionally, it explains how to authenticate with Azure, create resource groups, and configure Kubernetes and CircleCI for seamless application deployment. The guide is designed to offer a thorough understanding of the deployment process, making it suitable for developers familiar with Docker, ASP.NET Core, and Azure environments.

This comprehensive tutorial explores the integration of CircleCI runners with Scaleway's cloud ecosystem to efficiently manage AI/ML workflows, utilizing CI/CD practices and scalable GPU instances. By leveraging CircleCI's automation and ephemeral runners, the tutorial demonstrates cost-effective ways to provision, train, and deploy models while maintaining infrastructure flexibility. It emphasizes the benefits of using Pulumi for Infrastructure as Code to streamline resource provisioning and highlights the importance of configuring environments with the appropriate secrets and environment variables. The guide also covers practical steps for setting up models on scalable cloud infrastructure, including Docker installation for model serving, and the orchestration of AI/ML jobs using CircleCI's pipeline. The tutorial concludes with the importance of cleaning up resources to maintain cost efficiency and outlines how these practices can be applied to other cloud environments beyond Scaleway.

The text discusses the limitations of the traditional GitOps approach in software development, particularly in the context of modern applications that rely heavily on external changes such as third-party libraries, open-source tools, and microservices. The introduction of AI complicates the situation further, making it challenging to track changes and ensure software reliability and availability. The text argues that traditional CI/CD practices, which often rely on version control systems, are insufficient in this new landscape, emphasizing the need for more flexible and robust tools like CircleCI. CircleCI's platform is highlighted as being agnostic to version control systems, allowing it to better manage the complexity of software development at scale. The article stresses the importance of adopting best-in-class CI/CD solutions to effectively handle the evolving demands of software delivery and maintain developer confidence in a rapidly changing environment.

In this detailed exploration of MLOps automation, the article illustrates how CircleCI's inbound webhooks can be utilized to trigger workflows based on events outside of a code repository, enhancing the traditional CI/CD pipeline used in DevOps. By demonstrating the setup for automating model evaluations in a CircleCI GPU environment, particularly using an image-detection model trained to detect mask-wearing, it emphasizes the benefits of automating the testing process every time a model is updated on platforms like Hugging Face. This approach, facilitated by webhooks, allows for faster feedback loops and more efficient model delivery by eliminating the need for manual pipeline triggers. The article concludes by highlighting the evolving nature of software delivery practices and encourages users to leverage CircleCI’s webhook capabilities for more flexible and responsive CI/CD pipeline management.

The text provides a comprehensive guide on deploying a Node.js application onto an Amazon EC2 Linux instance and automating the deployment process using CircleCI. It begins by highlighting the benefits of AWS services like Elastic Beanstalk and EC2, catering to different team preferences, and then delves into the step-by-step process of setting up an EC2 server, configuring security settings, and installing necessary software packages like Node.js, npm, and pm2 for application management. The tutorial emphasizes the creation of a bash script for handling deployment on the EC2 instance and outlines the integration of GitHub and CircleCI to automate the deployment pipeline. It also offers security advice, suggesting the use of a reverse proxy server and a non-sudo user for SSH connections in a production setting, and provides links to similar posts for further learning.

The tutorial provides a comprehensive guide on building a continuous delivery (CD) pipeline for deploying large language model (LLM) applications using CircleCI. It extends a previous tutorial that covered continuous integration (CI) for testing LLM applications, now focusing on automating deployment to production environments. The process involves running nightly evaluations, allowing for human review of automated test results, and deploying the application after approval. The tutorial emphasizes the importance of monitoring real-world interactions to refine tests, and it demonstrates deploying a Flask server as part of the pipeline. By integrating LLM evaluations into the CI/CD pipeline, developers can streamline the deployment process, reduce manual intervention, and enhance responsiveness to user needs.

The tutorial explores the process of building and automating custom language model-powered applications using tools like LangChain and LangSmith within a continuous integration and continuous delivery (CI/CD) framework. It highlights the challenges of using large language models (LLMs) for enterprise applications and addresses how retrieval augmented generation (RAG) can enhance model accuracy by supplying relevant data at execution time. The guide demonstrates setting up a CircleCI pipeline to automate testing and evaluation, particularly focusing on the integration of LangChain's RAG application with LangSmith for debugging and testing. The example provided includes creating an automated question-answering chatbot that leverages LangChain’s functionalities to efficiently manage and test LLM applications. The tutorial emphasizes the importance of standardizing and automating workflows to improve the scalability and reliability of AI development processes, offering a detailed walkthrough for setting up the necessary environment, tools, and tests to achieve these goals.

Incompatible hardware often leads to application failures for distributed teams, prompting many to use containerization tools like Docker. This article outlines how to automate Docker image deployment efficiently by integrating CircleCI and Microsoft Azure to build a CI/CD pipeline for a Dockerized Spring Boot project. It requires basic knowledge of Java and Spring Boot, alongside accounts on CircleCI, GitHub, and Microsoft Azure, as well as updated installations of JDK and Docker. The process involves cloning a sample project, Dockerizing it, and deploying it to Azure by creating a Docker image and pushing it to the Azure Container Registry, followed by setting up an Azure Web App. Continuous deployment is enabled to ensure the web app is updated with each Docker image update. Finally, the article guides through connecting the project to CircleCI and automating the testing and deployment processes, enhancing the efficiency of application management by allowing the team to focus more on development rather than infrastructure.

Generative artificial intelligence (AI) is transforming software development by creating new content across various domains, such as text, code, images, and music. Tools like GitHub’s Copilot and OpenAI’s ChatGPT enhance development efficiency by automating routine tasks and offering real-time code suggestions, thereby accelerating development cycles and fostering innovation. However, the integration of generative AI poses risks, including potential copyright infringement, security vulnerabilities, and ethical concerns due to biases in training data. To harness AI's benefits while mitigating risks, developers must implement rigorous testing and validation processes, including code reviews, automated testing, and security analysis. Continuous oversight and adherence to coding standards are essential to ensure AI-generated code is secure, compliant, and of high quality. By incorporating these practices into continuous integration and continuous delivery (CI/CD) pipelines, organizations can safely leverage generative AI to enhance software development.