Muhamad Gulshair

Muhamad GulshairSep 10, 2026

The age of AGI is upon us, but centering a div is still hard. Found this bug on a Pixel 7A today — layout breaking on a screen size that "shouldn't" be an edge case. Could be device-specific, sure. But it's a good reminder of something we forget in all the AI hype: building software was never just about writing code that works on your machine. Real systems mean responsive design across screen sizes, cross-browser consistency, auth that's actually secure, and infrastructure that scales. It's a lot of unglamorous, detail-heavy work — the kind that doesn't show up in a demo video. So here's the question I keep coming back to: if AI still can't fully build a working, responsive, secure product without a technically skilled person steering it — how does the picture actually change in the next few years? I haven't even tried the premium Claude models yet, and I'm already hearing plenty of noise about engineers losing jobs to AI. Maybe that's coming. Maybe it's overstated. Either way, feels like a good time to actually pay attention instead of just reacting. Curious what other engineers are seeing — is this genuinely different from past "here comes the disruption" waves, or are we still early? #SoftwareEngineering #FrontendDevelopment #AI #WebDevelopment #instagram

Muhamad GulshairJan 29, 2026

Why LLMs Can’t Replace Developers (Yet) Over the last few years, we’ve seen a big shift in the tech industry: • Mass layoffs • Companies marketing “AI instead of humans” • High-profile cases like Builder.ai, where “AI” turned out to be just "Another Indian" Let me be clear: this is not an anti-AI post. Large Language Models are genuinely powerful. They can: Generate code at a senior-developer level Recognize patterns quickly Apply known techniques with impressive speed For certain problems, they’re incredibly effective. So where’s the real limitation? 👉 Context and ownership. If you’re building a simple 4–5 page portfolio website, an LLM or AI agent can handle it beautifully.A task that might take 2–3 days can be done in hours. That’s a real productivity win. Now compare that to a real production system with: 200–300 pages Multiple user roles with different permissions Complex forms evolving over weeks Workflow-driven features on top of heavy libraries Multiple developers working in parallel In that environment, a single form can outlive the context window of one LLM session. And this is where human developers still matter most. Building real software isn’t just about writing code. It’s about: Planning system architecture ahead of time Deciding what to build first and why Managing branches, PRs, and merges Knowing what needs refinement vs. what’s “good enough” Handling last-minute client changes that impact the foundation These require long-term memory, judgment, and accountability. AI struggles here. By default, it produces answers, not systems. It doesn’t naturally optimize for maintainability, scalability, or long-term trade-offs unless explicitly guided. As my friend Danish Riaz often says: “An LLM can only be as good as the person using it.” A strong developer knows how to: Ask the right questions Choose the right abstractions Guide AI outputs into an existing architecture A weak foundation doesn’t magically become strong just by adding AI. That doesn’t make LLMs bad—quite the opposite. They’re like extremely powerful libraries that remove friction and help developers ship faster. “Vibe coding” can be fun and useful. Skipping fundamentals is not. Software still requires: Thinking ahead Handling edge cases Making trade-offs with future impact in mind That’s something current LLMs can’t do independently. And yes—this post itself was refined with the help of an LLM. I brain-dumped the ideas, and it helped make them presentable. That’s the right model: AI as a force multiplier for developers, not a replacement. 💻🤝🤖 #ArtificialIntelligence #GenerativeAI #SoftwareDevelopment #Developers #AIinTech #SystemDesign #TechIndustry

Muhamad GulshairDec 16, 2022

Excited to announce I’m #OpenForBusiness and providing services on LinkedIn. Check out my services page for Search Engine Marketing (SEM), Advertising, Data Entry, Android Development, Web Development, Digital Marketing, Cybersecurity and Illustration. #android #digitalmarketing #marketing #linkedin #cybersecurity #illustration