Ujith B’s Post

Most engineers are using AI wrong. They treat it like a code generator. Prompt in, code out, ship it. That's the shallow end of the pool. The real leverage is using AI as a thinking partner. Challenge your architecture before you build it. Poke holes in your assumptions. Walk through trade offs you haven't considered. Review your design like a principal engineer would. The code is the easy part. The thinking is where engineers earn their salary. Teams that figure this out will outship teams that don't. Not by a little. By a lot. Link in comments. #ai #java #softwaredevelopment #coding

Agentic coding should be thought off as a tool to think and assist, not as a way to outsource product development to AI Agents. It may feel like a god send at first, but what it secretly brings is a disconnect that is a direct axe on human technical skill and development control. So how should a smart developer go about using AI? - Use Agentic AI tools to prototype a solution - Plan the software development pipeline using AI - Montior, Modify and Review code at every iteration. “Pair programming with AI” instead of “Administration of the Work done by AI” Treat the process like a team effort! 💡Insightful post Nelson Djalo #agenticai #softwaredevelopment #vibecoding

The hardest part is ensuring you are taking the extra time to prompt AI in this manner. It’s so easy to prompt LLMs in a way that gives you the answer.

I used AI in backend development for 30 days. Here’s the truth. AI is powerful—but only in the right areas. Where it helps: Boilerplate code (huge time saver) Debugging hints (helps you move faster) SQL queries (quick and mostly accurate) Documentation (fast and easy) Where it doesn’t: System design decisions Scaling and performance trade-offs Production debugging The takeaway: AI won’t replace backend developers. But developers who use AI well will outperform those who don’t. Are you actually using AI in your workflow—or just experimenting? #AI #BackendDevelopment #SoftwareEngineering #AICoding #Developers #Programming #SystemDesign #SQL #DevTools #Tech

When we use AI, prompt engineering is very important because it determines how well the AI understands what you’re saying. As I understand it in simple terms, prompt engineering is about how you put your thinking into words so that AI can understand and give you better results. Most of us only use 2–3 layers in our prompts, and that’s why the output often feels generic. Here are the 6 layers of prompts I’ve started applying in my daily work. You don’t always need to use all of them; apply them as needed to get better results. The 6 Layers 1. Role 2. Context 3. Task 4. Format 5. Constraints 6. Examples Example Role: You are a... Context: I’m working on... Task: Write/Create/Generate... Format: Output should be... Constraints: Do not... Example: Here’s a reference: [paste example] 6-layer prompt: Role: You are a senior backend engineer experienced in high-scale systems Context: I’m working on a Node.js API where a single endpoint fetches user orders with multiple joins and is taking 2–3 seconds response time Task: Analyze and suggest improvements to optimize query performance and reduce response time Format: Provide step-by-step optimization suggestions with code examples where needed Constraints: Do not suggest caching as the first solution; focus on query optimization and indexing first Example: Current query uses Sequelize with multiple includes and filtering on userId #SoftwareEngineering #Developers #Coding #Programming #Tech

Unpopular opinion: AI is making a lot of developers faster. But not better under pressure. They can ship code. They can explain patterns. They can generate tests. They can clean up boilerplate. But when production gets weird, speed stops mattering. That’s when engineering depth shows up. Can they trace a failure across services? Can they spot retry amplification? Can they question a timeout budget? Can they understand why a healthy service is still part of a broken request path? That’s the gap I keep thinking about. AI is raising coding speed. But it may also be hiding how few engineers truly understand production behavior. Debate: What creates stronger engineers in the long run? A) shipping fast B) debugging real production issues C) mastering system design D) writing more code My vote: B first. What’s yours? #Java #AI #BackendEngineering #DistributedSystems #SpringBoot

I’ve been integrating AI into my daily workflow for a while now, and the realization hit me: AI has effectively solved the "coding" problem. I can now deliver features and fix bugs without typing a single line of manual code. As someone who loves the craft of writing code, it was a shock. But it also brought a deeper clarity. While AI is world-class at coding, it is indifferent to the problem. It doesn’t understand the root issues impacting your users, nor can it navigate the nuances of a complex, long-term architecture. So, how do we stay effective as Software Engineers? We have to level up. If we don’t, we’re just generating a mountain of automated code that will inevitably lead to massive maintenance costs and "hallucinated" bugs. I believe the path forward is clear: Engineers must move closer to the Product. Our role is no longer typing code. It’s: 1- Defining the problem 2- Shaping the solution 3- Ensuring the system holds over time

I used to spend 3 hours writing code. Now it takes 20 minutes. Not because I got smarter. But because I stopped being afraid to use AI as my co-pilot. Here's what changed for me as a software engineer: Before AI tools: → Stack Overflow tabs open in 10 windows → Debugging the same error for 2 hours → Writing boilerplate code from scratch... every single time After using AI tools: → Instant context on bugs I've never seen before → First draft of functions in seconds → More time thinking about architecture and logic But here's the real talk — AI doesn't replace your thinking. It amplifies it. The engineers who will thrive in the next 5 years are not the ones who avoid AI. They're the ones who learn to work with it better than anyone else. I'm still learning. Still experimenting. Still making mistakes. But I'm also shipping faster, learning more, and enjoying the process more than I ever did before. If you're a dev who hasn't explored AI tools yet — what's stopping you? Drop your answer in the comments. I'm genuinely curious. #SoftwareEngineering #AI #TechCareers #ArtificialIntelligence #DeveloperLife #AITools #CodingLife #FutureOfWork #LinkedInTech

We’ve all seen this meme, and let’s be honest… we’ve all met someone who acts a little too much like the guy in the second panel. 😂 AI tools are incredible for boosting productivity, but they are exactly that: tools. They won't replace a deep understanding of your chosen languages, frameworks, and system architecture. Real software engineering is about problem-solving, not just prompt engineering. #SoftwareEngineering #TechCommunity #Coding #AI

𝗔𝗜 𝗱𝗶𝗱𝗻’𝘁 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗲 𝘁𝗵𝗲 𝗷𝗼𝗯. 𝗜𝘁 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱 𝘁𝗵𝗲 "𝗱𝗼𝗶𝗻𝗴" 𝘀𝗼 𝘄𝗲 𝗰𝗼𝘂𝗹𝗱 𝗳𝗼𝗰𝘂𝘀 𝗼𝗻 𝘁𝗵𝗲 "𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴." A lot of people think AI makes engineering "easy." In reality, it has made the need for senior-level thinking more intense than ever. In my daily work as a Lead, I’ve seen my time shift:   𝗕𝗲𝗳𝗼𝗿𝗲: Hours spent writing boilerplate, manual syntax debugging, and slow iteration.   𝗡𝗼𝘄: Rapid prototyping, instant idea validation, and a focus on high-level system decisions. 𝗧𝗵𝗲 𝗿𝗲𝗮𝗹 𝘀𝗲𝗰𝗿𝗲𝘁 𝘁𝗼 𝗵𝗶𝗴𝗵-𝗾𝘂𝗮𝗹𝗶𝘁𝘆 𝗔𝗜 𝗼𝘂𝘁𝗽𝘂𝘁? 𝗔𝗻 𝗶𝗺𝗽𝗹𝗲𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻-𝗳𝗶𝗿𝘀𝘁 𝗺𝗶𝗻𝗱𝘀𝗲𝘁. Months ago, I was frustrated by AI giving unrelated or incorrect answers. I found a simple fix: I started adding a clause to every prompt: “𝗔𝘀𝗸 𝗺𝗲 𝗮𝗹𝗹 𝗰𝗹𝗮𝗿𝗶𝗳𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻𝘀 𝗯𝗲𝗳𝗼𝗿𝗲 𝘆𝗼𝘂 𝗽𝗿𝗼𝗰𝗲𝗲𝗱.” I remember the first time I did this with Grok—it came back with so many questions I felt overwhelmed. Why? Because answering those questions requires the one thing AI can’t do: 𝗗𝗲𝗲𝗽, 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝗶𝗰 𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴. Today, my workflow with agents like Cursor and Claude doesn't start with code. It starts with an implementation_plan.md. I spend more time thinking through the logic and documenting the plan than I do actually generating the code. 𝗧𝗵𝗲 𝗟𝗲𝘀𝘀𝗼𝗻: In 2026, good thinking equals better AI results. AI hasn't reduced the need for skill; it has increased the premium on a Senior Engineer’s ability to architect and plan. 𝗧𝗼 𝗺𝘆 𝗳𝗲𝗹𝗹𝗼𝘄 𝗟𝗲𝗮𝗱𝘀: How has your "thinking-to-coding" ratio changed lately? Are you spending more time in Markdown files than in Python or Rust? #EngineeringLeadership #AIEngineering #SystemDesign #SoftwareArchitecture #CloudNative #DevOps #GCP #CursorAI #CleanCode