Omar Fouda

#SoftwareEngineering I find it fascinating that in the US, something is called a tri-angle (three angles) whereas in Germany, the same thing is called Drei-Eck (three corners). This illustrates the difficulty of talking about the same thing, if you have very different conceptions about it. This relates to software engineering in several ways. The difficulty of (corners vs. angles) 1. Naming things 2. Unified language in DDD 3. What is core to a feature? (talking to PMs) 4. How to architect things 5. What to abstract over (corners vs. angles) Taking 1: What I experienced in large code bases, with generations of developers working on it, often the initial developers had left, the same thing is named differently in different parts. e.g. the database calls something "User", the code calls the same thing "Customer". This is the core of the idea of the unified language (2) of DDD. People look at things and focus on different aspects by naming it.

180

127 Kommentare

lately I've been focusing more on the engineering management track, and one of the most common cases that I see with software engineers who struggle with there progression in there career is #communication and by this I mean how skilled you are in communicating technichal problems in to business language as your mind is used to think in terms of code, classes, packages, design, technichal debt, etc.. your mouth should speek business, vision, OKRs, where do we stand from our objective, etc.. from my point of view ... this is a must have skill for any senior engineer who is planning to progress more in his career ... if not the most important one

141

10 Kommentare

🚀 Freelance vs Full-time Software Development in Germany - When to choose what In Germany's technology market, both freelance developers and permanent employees play important roles. The right choice depends on your business needs. ✅ Freelancers are the better choice when ⚡You need very specific expertise for a clearly defined project or technology stack ⚡Speed is important, since freelancers can usually start quickly without long hiring processes ⚡Your workload is project-based and does not justify a permanent hire ⚡You want flexibility to scale teams up or down without long-term obligations 👩‍💻 Permanent employees are the better choice when ⚡You want to build long-term product knowledge and strong company culture ⚡Continuity in core systems is essential ⚡The role requires deep integration with the business and its processes In short: 👉 Choose a freelancer for focus, speed, and flexibility 👉 Choose a permanent employee for stability, culture, and continuity The value lies in knowing which model fits your current challenge. #Freelance #SoftwareDevelopment #TechHiring #ITFreelance

38

I was misunderstood. I gave a talk about a startup in which I had to work with some junior-level software developers. I tried to emphasize that, in a startup, you often have to adapt your strategies to the real circumstances that you face. If you can afford to spend $250,000 and hire senior-level, genius-level software developers, that is awesome, but often early-stage startups have limited funds and they end up paying very junior-level software developers $35 an hour. And so you have to adapt your strategies to the reality that you are working with very junior software developers, and you have to cater to the fact that they do not know crucial facts about software development. But some people in the audience completely misunderstood me. They felt I was criticizing the junior-level software developers. They felt I was, in essence, saying that the junior-level software developers were stupid. When I listed all of the things that the junior-level software developers did not know, some people felt that I was mocking them. But my point was that the lack of knowledge was a reality that the startup needed adjust to, because the startup didn't have the money to hire experienced software developers who might have known more. I need to rethink how I communicate this idea, because I think it is something important for the leadership to think about, but the way I phrased it offended several people.

43

54 Kommentare

Yesterday I cleaned up about 60% of code generated by an agent that prefers to stay unnamed :). It brought a pile of unnecessary complexity. Again. And I caught myself thinking that we might be losing the battle against endlessly growing complexity with the help of AI. Software development is a continuous fight with complexity. We have been trying to cage it with boundaries: managed services, frameworks, libraries, microservices. We argue about architectures to keep systems maintainable. Layered, hexagonal, clean, domain-driven, event-driven, CQRS, event sourcing, SOA, micro frontends. We do it with a single goal: to minimize the blast radius and introduce clear contracts. AI flips the table. Generation is easy. It ships code with no effort and ignores trade-offs. Maintenance is hard. Coupling, cohesion, run costs, reliability — all your problem. Think of somebody gifted you a dog. Zero purchase price, daily work for decades. Feeding, training, vet bills, walks — all yours. Enjoy! It feels like empowered with AI we’re about to lose this fight against the uncontrollable growth of entropy in source code. When engineers hit the ceiling and pull back: they pay tech debt, they refactor, they extract libs, they even say no to features. They deal with it. AI hits the ceiling and keeps going, generating layer after layer of code. Crap over crap.

1.467

163 Kommentare

🚀 Why I believe n8n will become the new norm of software development! Jan Oberhauser regularly gives the example of knowing Excel was one of the most powerful skills during the early days of the computer revolution. He thinks we’re at that same moment again, only this time, it’s n8n, the new Excel. 💡 In this short video (42 seconds), I’m doing a refactor of the Execution Data node. I move it from one place to another because I need the data from the execution in multiple parts of the workflow. All I do is delete a few links, drag the node, snap the arrows back and boom, the data instantly becomes available across the entire workflow, all the way to the rightmost part of the canvas. This is magical. 🔥 That single move changed the scope, the way variables propagate, and how the Execution Data node interacts with the rest of the workflow, all visually, in real time. No redeploys. No broken code. No headaches. If I had tried to do the same thing with LangChain or by writing traditional code, it would’ve taken me hours, not seconds. After 17 years as a full-stack engineer, I’ve never felt this level of productivity before. 🧠 n8n + AI-assisted development (aka “vibe coding”) = pure magic. This tool makes you a 10x developer, not because it replaces your skills, but because it amplifies them. If you haven’t tried n8n yet, do yourself a favor: go build something small today. You can start from the official n8n Starter Session tutorial series from the first comment. You’ll see what I mean. #n8n #automation #nocode #AI #developers #productivity #workflowautomation

668

61 Kommentare

I’m attending PromCon 2025 this week in Munich. It’s great to see the #Prometheus community come together again. Day 1 was packed with great sessions, but one talk that really caught my attention was by Julius Volz, co-founder of Prometheus: “Why I Recommend Native Prometheus Instrumentation over OpenTelemetry.” In a time where OpenTelemetry (OTel) dominates most observability discussions, Julius offered a clear, experience-based argument for why native Prometheus instrumentation remains the better option when Prometheus is your main monitoring system. A few key insights from his talk 👇 🔹 Prometheus is a monitoring system, not just a metrics pipeline. It discovers targets, actively scrapes them, and automatically records their health via the up metric. With OTel’s push model, that visibility is lost, Prometheus can’t tell if a service has stopped reporting or disappeared. 🔹 Performance and simplicity matter. He shared benchmarks showing the Prometheus Go client performing up to 26× faster (and even 53× faster with cached labels) than the OTel SDK in common counter operations. Prometheus libraries are small, focused, and optimized for high-throughput systems. 🔹 Metric naming consistency. While Prometheus 3.0 adds UTF-8 support, OTel’s use of dots and dashes in metric names still leads to translation issues and awkward PromQL syntax. Native Prometheus metrics remain clean, predictable, and fully compatible. 🔹 Simplicity over complexity. Prometheus’s text-based exposition format can be implemented anywhere, even with a few lines of Bash, without additional protocols or SDK layers. It stays true to the Unix philosophy: simple, transparent, and efficient. Prometheus itself is an open standard with wide adoption, mature tooling, and a strong community. Choosing native instrumentation isn’t about resisting change, it’s about using the right tool for the right job. So the first day of PromCon 2025 delivered plenty of valuable discussions around metrics, scalability, and the future of observability. Looking forward to the rest of the conference. #PromCon #Prometheus #OpenTelemetry #Observability #Kubernetes #DevOps #Monitoring

23

1 Kommentar