Konstantin Kustau’s Post

Should CS programs be teaching orchestration with AI tools? Mihail Eric Head of Ai at Monico says yes in this article. What do you think?

Engineering teams everywhere face the same bottleneck: not enough developers. Hiring takes months. Good contractors cost $150+/hour. And by the time someone is onboarded, the sprint is already behind. Elite Coders takes a completely different approach. Instead of hiring humans or spinning up anonymous AI tools, you get AI-powered full-stack developers at $2500/month. Here is what makes it different from other AI coding tools: - Each dev gets their own name, email, avatar, and personality - They integrate directly with your Slack, GitHub, and Jira - They pick up tickets, write code, push PRs, and respond to code reviews - Available 24/7 with zero ramp-up time They don't just generate code in isolation. They show up on day one, join your workflows, and start shipping like a mid-level engineer who already knows your stack. Built for startup founders, CTOs, and engineering managers who need more development capacity without the 3-month hiring cycle. 7-day free trial, no credit card required. Check it out: https://www.elitecoders.ai

Claude Skills is so awesome! You can get it to take over those small non-tech tasks that fill your day, so that you can focus on the strategic stuff. It's gonna revolutionize non-tech jobs! That's what they say. Meanwhile, the reality: PM: I reviewed your last tasks for completion. Here are 3 small things that are missing from the defined scope. Can you please add those? Dev: Sure. Let me check. [Checks the detailed descriptions. Reviews the code. Double-checks the original ticket. None of the "things missing" seems to have been in the specs. Two are clearly hallucinated. The third one might have been inferred. Maybe. Maybe not. Probably not. Not really. More than an hour has passed.] Dev: Are you sure these *were* parts of the scope? PM: Let me check... Here's the thing. The review (or I should say the "review") was run by Claude Skills. "Take the task description. Check the PR for completion. Report any missing parts." Then the overeager Claude has gone over its head and hallucinated stuff. A PM has just saved themselves 15 minutes of a quick functional test of the app. A developer invested 5 times that in investigating a hallucination case. A classic case of work slop. One could bet how long it will take a developer, learning by example, to create their own Claude Skill to respond to PM's remarks... "If you can't be bothered to do your work well, why should I be bothered to do mine any better?" Aside from the sociotechnical part, there's another aspect to this situation. We have specific, verifiable rules for code (syntax, coding standards, principles, etc.); we don't have nearly as much rigorous scrutiny for plain English. In other words, for LLM, it's easier to hallucinate in English than it is in TypeScript. As coding has become one of the main applications of LLMs—and one of the reliable revenue sources for AI labs—there's a lot of focus on getting it "right." More general contexts? Not so much. I mentioned it before: There are not enough casual $20-a-month users for LLMs to make AI labs profitable. There are not enough developers on planet Earth to do so either. They need to push for other business applications. And Claude Skills is precisely that. That's one reason why Anthropic's story is so interesting to observe: https://lnkd.in/dFM7gkH5 However, the combination of these observations only means more work slop. We keep discussing whether AI will be able to generate software products, thinking in terms of code and code only. In reality, it may as well be local optimizations that create systemic inefficiencies. Systems Thinking 101. And that's all, assuming a positive scenario, where work-slop receivers don't start playing the same game, and get the very same Claude Skills to automate away the slop. Creating even more of it. Except for someone else. What does the endgame look like, again?

The floor just dropped out for Junior Developers. 🕳️ And most of them don't see it yet. A Senior engineer with AI can now do in seconds what used to take a whole Junior + Senior team days. ⚡ That's not hype. That's the new baseline. Here's the brutal truth about what's happening: 🤖 AI handles the boilerplate, the unit tests, the "glue code" 📉 Companies now want Mid-level skills at Junior pay 🚪 The entry door didn't just get smaller — it moved floors But here's what nobody is saying loudly enough. 👇 "Just prompting" will get you hired. It won't keep you employed. 💀 When the AI hallucinates a subtle architectural flaw, the developer who skipped the fundamentals won't catch it. Until it crashes in production. 🔥 So if you're early in your career, here's what actually matters now: 🧱 Learn Data Structures and Algorithms — the AI uses them, you need to audit them 🚀 Build full systems end-to-end, not just isolated functions 🔍 Explain every line the AI writes — if you can't explain it, you don't own it 🎯 Be the human in the loop, not the human who outsourced their brain Seniority isn't about how many languages you know anymore. It's about your ability to verify what the AI tells you is true. 🧠 The Junior Developer isn't dead. But the one who coasts on AI without building fundamentals? 💀 That one is. Are you a Junior dev navigating this shift right now — or a Senior watching it happen on your team? Drop your perspective below. 👇 #JuniorDeveloper #AIEngineering #SoftwareDevelopment #TechCareers

I keep seeing the same question everywhere, “Will AI replace junior developers?” Here’s my honest take, as someone who just finished a coding bootcamp and is entering this market right now: No, I don’t think AI will erase junior developers, but it will absolutely redefine what “junior” means. Throughout my training, I used AI to troubleshoot unfamiliar error messages, test ideas, and catch syntax issues quickly. It made me faster and more efficient, but it did not make me unnecessary. To me, being a junior developer today is not just about learning how to code. It is about learning how to use AI effectively, question its output, and recognize when something that looks right is actually wrong. The bar is shifting from simply producing code to thinking critically, learning quickly, and understanding the bigger picture behind the work. Let me give examples: What AI can’t do: ❌ Understand what the client actually wants versus what they said they want ❌ Debug why your specific architecture made a weird decision three sprints ago ❌ Sit in a standup and translate between the business team and the dev team ❌ Build relationships with a team What I bring that pairs well with AI: ✅ The ability to communicate with both clients and coworkers to understand problems and troubleshoot coding issues ✅ Creative problem-solving from a game design background ✅ I know how to learn fast and document what I learn ✅ A mindset of using AI to support the work, not replace the thinking behind it And I believe, there are still a lot of good reasons to hire junior developers. Every strong senior developer was once given a chance to learn. Juniors developers bring enthusiasm, a fresh perspective, adaptability, and a willingness to learn quickly. They grow with the team, absorb the way a company works, and can become some of the most valuable long-term assets you have if they are trained well. My opinion is the developers who thrive alongside AI won’t be the ones who ignore it, they’ll be the ones who know when to use it and when to trust their own judgment. I’m actively job hunting right now. If you’re building a team and want someone who uses AI as a tool, not a crutch, let’s connect. What’s your take, is AI an asset or threat for junior devs? 👇 #artificialintelligence #juniordev  #webdevelopment #softwaredevelopment  #aitools #techcareers #opentowork #github

"The job market is brutal for junior devs." I see this everywhere. What I don't see? Anyone telling them what to actually do about it. Here's the truth nobody's saying: AI is the biggest opportunity juniors have ever had. → No more endless tutorials → No more hunting for resources → No more waiting months to understand a concept Everything you need to know is available. And you can apply it faster than ever before. The era of hand-coding is almost over. Senior devs are openly saying they haven't written a single line of code in weeks. So what matters now? Not typing code. Understanding systems. Here's what I'd recommend: 1. Go deep on your main language Don't just use it. Understand how it works under the hood. Every fuzzy concept? Dig until it clicks. 2. Learn what makes apps scale How do you go from 5K users to 5M? Architecture matters more than syntax. 3. Master the fundamentals • Design patterns • SOLID, DRY, KISS • TDD, DDD • When to use microservices (and when not to) • CAP theorem, CQRS 4. Context is everything Your tech choices depend on business goals. Learn to think in tradeoffs, not absolutes. 5. Practice out loud Code reviews. Discussions with AI. Explain concepts verbally. If you can't explain it, you don't understand it. You have a 24/7 mentor that knows more than most seniors. Use it. Apply concepts daily. Build projects that demonstrate understanding. Learn in months what used to take years. Juniors who get this will leapfrog seniors who don't. Take your chance.

Some disjointed thoughts on how AI has changed the landscape of software engineering. The Software Engineer In my opinion, the "software engineer" has been almost redefined - the impact of interviewing a candidate with a project-based task and seeing how they plan it out, predict possible blockers, and bring up concerns within the tools used is so much more valuable now than "hey do this LeetCode" or even "solve this algorithm problem". This is now true across all levels, even for an Intern or Junior. To me, if we were to use AI in the most efficient way for development, every software engineer is becoming something of a product manager. Eden Ding and I have been having conversations on how to approach hiring as well - do we allocate for more engineers or hire less engineers and focus on taking advantage of agentic coding instead? Do we hire a bunch of coachable juniors or just a few seniors? Would love to hear other people's thoughts on this as well! Shiny Object Syndrome I genuinely believe that when you're building a product with any sense of urgency in today's AI-driven world, success is defined by finding the fine line between being a canary that scrolls hackernews and relevant Github repos for the best "new way" to build (it's an addiction) and actually sitting down and building it. The Agentic Employee Recently, my team and I at NationGraph introduced a coding agent that was fully integrated within our Linear, Github and Slack workspaces. After a fun session bullying the poor agent, we started using it to boost our productivity. The agent is able to create/enrich linear tickets, take on linear tickets itself, even put PR's in our repos and ping us for review with full E2E self-testing - all of the elements of a Junior Engineer if not Intermediate Engineer. It was a part of our Sprint Planning and it was checking off some tasks as quickly as we can write them. Even for me, this felt really weird and at a certain point Adam Omarali was telling me about how he doesn't want to have it do everything because he'd rather the code it himself than PR review it, even if the Agent does it perfectly. There's a lot to think about with how we'll be approaching agentic development in the coming months. Companies such as Stripe have thousands of PRs closed each cycle from just purely E2E AI agents (See: https://lnkd.in/eg9C6pfR), and I think we're all eventually going to converge on this framework but who knows! Thanks for listening ya'll, and lets keep building!

The industry is starving for junior developers. But here’s the paradox. A “junior” today is basically a mid-level from 3 years ago. AI raised the baseline. Tools accelerated learning. Expectations exploded. Companies now expect juniors to: • ship features • understand production code • use AI effectively • learn fast But there’s something else people ignore. Juniors still have what the industry desperately needs: • curiosity • initiative • hunger to prove themselves • ideas shaped by the current market Their eyes are still on fire. They care. They experiment. They try things seniors stopped trying years ago. Meanwhile many “seniors” quietly stopped evolving. They didn’t become architects. They didn’t become CTOs. They didn’t become managers. They just kept the title. And after 10–12 years in the industry some of them are simply… tired. Burned out. Comfortable. Playing it safe. So sometimes the person pushing the product forward is not the burned-out senior with a decade of experience. It’s the junior who started 18 months ago and still believes everything is possible. No juniors today → no seniors tomorrow. The industry isn’t drowning in juniors. It’s starving for them. And if a “junior” with 18 months of experience can already replace you… maybe the problem isn’t the junior.

Most developers are not losing jobs because of AI. They’re losing jobs because they only know how to code… not how to solve problems. Today, companies don’t care if you know: React, Next.js, or 10 frameworks. They care about: → Can you build something end-to-end? → Can you ship fast? → Can you think like a product owner? AI didn’t replace developers. It exposed the difference between: Code writers vs Problem solvers The ones who win in 2026 are not the best coders… They are the fastest learners + builders. What do you think — Is AI replacing developers or upgrading the game? #AI #ArtificialIntelligence #Developers #DeveloperLife #FrontendDeveloper #FullStackDeveloper #SoftwareEngineer #Programming #CodingLife #TechCareers #CareerGrowth #FutureOfWork #TechIndustry #BuildInPublic #StartupLife #RemoteWork #OpenToWork #Hiring #TechHiring #Innovation #DigitalTransformation #ProductDevelopment #LearnToCode #100DaysOfCode #DevCommunity #WebDevelopment #ReactJS #NextJS #TypeScript #JavaScript #SaaS #Automation #AIinTech #MachineLearning #ProductMindset #Entrepreneurship #GrowthMindset #ProfessionalGrowth #Opportunities #Networking

The developers who will win in 2026 aren’t the ones who know *everything*… They’re the ones who know the **right skills**. Right now, the game is shifting fast ⚡ Here’s what’s actually trending (and paying): 🔥 AI Integration > Just Coding Anyone can write code now… But can you build with AI tools, APIs, and automation? 🌐 Full-Stack Thinking > Single Skill Frontend + Backend + Basics of DevOps = 🔥 ⚡ Performance & Speed Optimization Slow apps are dead apps. Period. 🔗 API & System Integration Everything is connected — apps, payments, CRMs, AI tools. 📱 Real-World Problem Solving Companies don’t want coders… They want people who can **solve business problems with tech**. The truth? Syntax can be learned in weeks… But **skills that create impact = real value**. So instead of chasing 10 languages… Master 2–3 **high-impact skills** deeply. Because in the next 2–3 years, average developers will struggle… but skilled developers will be in crazy demand 🚀 #Developers #TrendingSkills #AI #WebDevelopment #FullStack #TechCareers #Coding #FutureOfWork #LearnToCode #IndiaTech