Joyvis Santana

🚀 Duolingo proved that Kotlin Multiplatform (KMP) is a real advantage. Their incremental adoption reshaped how they build products: 🔁 Less duplicated code, faster releases. ⚙️ Projects like Math Grading and Maker cut development time from months to weeks. 🤝 A single shared logic base across Android, iOS, and backend. With KMP, Duolingo moved from small isolated modules to a “KMP-first” strategy, building a clean architecture and a shared state engine powering complex, interactive features like games and video calls. https://lnkd.in/dW3Q_Myk #KotlinMultiplatform #KMP #Kotlin #MobileDevelopment #Duolingo #AndroidDev #iOSDev

223

4 Comments

46 Essential API Architecture Styles for Modern Applications Want to build better, faster, and more scalable systems? Understanding API architecture styles is non-negotiable for modern developers and architects. Here’s a quick breakdown of a few key ones: GRPC: A high-performance, language-agnostic framework ideal for distributed systems. SOAP: XML-based protocol that ensures strict standards and security for enterprise-grade web services. GraphQL: A powerful query language that fetches exactly the data you need—no more, no less. Webhook: Event-driven magic for real-time updates and notifications. REST: The go-to HTTP-based API style that powers most of today’s web applications. WebSocket: Enables real-time, bidirectional communication—perfect for chat, gaming, and live streaming. Why it matters? Each architecture has its unique strengths, and the right choice can: Enhance performance Boost scalability Improve user experience Future-proof your applications Choose wisely. Build smart. Scale seamlessly. Have a favorite API style? Share it in the comments! Image Credit - Amigoscode #API #SystemDesign #WebDevelopment #TechArchitecture #GraphQL #RESTAPI #WebSocket #GRPC #Developers #SoftwareEngineering #TechTrends

99

38 Comments

8 rules for Software Engineers using Claude Code. I've been using Claude Code for months: And it's completely changed how I think about building software. Here's what I learned: 1. Claude file as project memory Create a CLAUDE {dot} md file with project rules, commands, and guidelines - it loads with every request and acts as Claude's persistent memory. 2. Test driven development (TDD) Have Claude write tests first, confirm they fail, then implement code to pass them. You don’t need to always do this but it's just one way to constrain and guide AI. 3. Write perfect commit messages Just type "commit" in Claude Code and it will handle git add and commit. The Claude commit messages will be the best you've ever seen in your career. 4. Have Claude review Claude Open two windows and have one review the other's work - perfect for code review iterations. 5. Throw away bad work Don't fight to fix derailed attempts - just git commit frequently and revert when Claude goes off track, then restart with better prompts. 6. Plan mode (Shift+Tab+Tab) A new feature that lets you have a conversation with Claude to generate a plan. 7. Slash commands Define custom slash commands in .claude/commands/ with interpolatable arguments - create commands for refactoring, linting, PR reviews. 8. Write PRDs not prompts The modern AI workflow is about giving more context to the models. You can write detailed PRDs for features you need as markdown files and give them to Claude to plan and implement. “Please implement section 1.5 of the PRD found here: FILE PATH” Thanks for reading. -- PS: If you want weekly insights and practical guidance on leveraging AI, I write a free newsletter here: https://leverageai.co/ PS: Enjoy this? ♻️ Repost it to your network and follow Owain Lewis for more.

522

89 Comments

What's new in Angular 20? For the first time, I explored each single change of this #angular release, diving into the codebase to understand its context. I learned a lot and created a huge article to share the insights. - experimental Vitest support - enhance dynamic component creation - new style guide - Signals new stable APIs - and an endless list of dx enhancements Discover the blog post in comments:

90

2 Comments

I keep seeing the same debate: should we use RabbitMQ or SQS? Kafka or a message broker? Most of the time, this discussion starts in the wrong place. We jump straight into tools instead of asking the more fundamental question: what communication model and what protocol does this architecture actually need? On one side, we have AMQP-style messaging. RabbitMQ, Azure Service Bus, and SQS/SNS give you a push-based, message-oriented model. The focus is on message exchange, routing, fan-out, dead-letter queues, and elastic horizontal scaling of consumers. This tends to shine inside a domain, integrating bounded contexts that are closely related and change frequently together. On the other side, we have message logs, such as Kafka and similar systems. These are pull-based, partitioned, and durable, designed for reprocessing and long-term history. They fit better when integrating different domains or business areas, where contracts are more stable, reactivity needs are lower, and teams are not changing things together every week. Seen from this angle, the question is less “RabbitMQ vs SQS vs Kafka” and more “AMQP-style messaging vs log-based streaming”. Only after you choose the right family of protocols and semantics does it really make sense to pick a specific product. In other words, architecture should start from principles and protocols, not from tools. The technology is an implementation detail of the communication model your domain actually needs. #EventDrivenArchitecture #SoftwareArchitecture #DistributedSystems

127

33 Comments

Real-time Node.js pipelines should not stall on outdated clients. → current options lag on maintenance, worker-thread support, and modern APIs → serialization boilerplate clutters code reviews In three days we’ll show you what happens when performance engineering meets developer-first design. Clear your calendars.

94

5 Comments

A staging environment is a safe replica of production where you can test features before going live. In this post, we'll learn how to create a staging environment in Ruby on Rails and deploy to it with Kamal. It also covers the basics of Rails environments and shows how to configure Kamal for staging deployments. As always, I hope you find it useful and you learn something new. https://lnkd.in/dewwVXma

44

6 Comments

💡 Event Storming: the most practical tool I’ve found for DDD When I first started working with Domain-Driven Design, one of the toughest questions was always: where are the boundaries? I’ve learned the hard way that if you jump into coding or system design before discovering those boundaries, you almost always end up refactoring later, and not in a fun way. That’s why I’ve come to rely on Event Storming. It’s the most practical way I’ve found to uncover how a business really works, where one domain stops, and another begins. In a typical session, we start with events, which are things that happened in the past that matter to the business. Then we add activities, which describe what people do, and roles, which show who performs them. Very often the boundaries appear around those roles, because they belong to different departments or own different responsibilities. That ownership is usually a natural indicator of where one domain ends and another begins. For example, in a project around learning platforms, events like Course Created, Student Enrolled, and Payment Processed showed clear boundaries. Teaching and learning were owned by educators and students, while billing belonged to the finance department. Looking at the events, activities, and the departments behind the roles made it obvious that these were separate domains with their own upstreams and downstreams. We didn’t need a big upfront design document because the business itself revealed the lines. What I like about Event Storming is that it is not just for developers. Product managers, architects, and domain experts all get something out of it. When everyone sees the same flow, the right boundaries become obvious. For me, Event Storming has moved from “a nice workshop” to an essential step before I commit to design or code. It saves time, aligns people, and gives me confidence that the system will reflect the business rather than just my assumptions. I especially recommend architects and senior developers to learn this method, because it will directly improve the way you design and build systems. #EventStorming #DDD #DomainDrivenDesign #SoftwareDevelopment #SoftwareEngineering

63

14 Comments

Modern tooling promises speed but how much time can we actually save, and can it happen without a single line of refactoring?   That question led Bernardo Oliveira to test tools like Rolldown, Sass-Embedded, Lightning CSS, and Oxlint across our micro-frontend environments. The results were surprising.   These drop-in replacements delivered real performance gains in both build and linting times with none of the usual complexity. No dramatic migrations, just smart swaps that help teams move faster and stay focused.   This isn’t a trend piece. It’s a data-backed look at what happens when engineering time is treated with care.   Read the full story on our blog: 👉 https://mb4.me/LF0shIkv.

67

4 Comments