Amlan Gupta

AWS CEO dismisses #AI #job loss fears, says Amazon plans to hire 11,000 interns in 2026 AWS's CEO, Matt Garman, said #Amazon sees demand for hiring new software engineers "accelerating." He said Amazon will hire 11,000 SDE interns in 2026 — the company says it's similar to recent years. Other AI leaders warn coding tools could significantly disrupt software engineering jobs. https://lnkd.in/gqVV7x7N?

104

2 Comments

I’ve been diving into how Vector embeddings works in Open Search. Instead of just reading about it, I decided to build a quick experiment to see Semantic Search in action in. The goal was simple: How can I use Vector Embeddings to navigate a complex Myntra dummy dataset based on "intent" rather than just matching keywords? 🚀 The Goal: Stop matching keywords, start matching "vibes." I hooked up all-mpnet-base-v2 (via Transformers) to an OpenSearch cluster to see if I could build a search bar that understands intent. If you search for "something for a rainy hike," it finds Waterproof Trekking Pants—even though the word "rainy" appears nowhere in the product description. The AI doesn't just match letters; it understands that "rainy" and "water-repellent" are neighbors in a 768-dimensional mathematical space. 📐 What's under the hood: 🔹 Embeddings: Turning text into vectors using NLP. 🔹 Vector DB: Using OpenSearch K-NN (K-Nearest Neighbors) for the heavy lifting. 🔹 The Fun Part: Realizing that "Waterproof" and "Rainy hike" are neighbors in a vector world. It's not a finished product—just an exploration into the power of high-dimensional search. Give it a spin here: 🔗 https://lnkd.in/gQsskiQe #AI #VectorSearch #OpenSearch #NLP #MachineLearning #WebDev #BuildInPublic #SoftwareEngineering

7

One year ago, I left the Bay Area, wrapped up nearly 8 years on US work visas (TN + H1B), and moved back home to Toronto. Living in the states as a skilled immigrant came with a constant hum of uncertainty around visas, but also lead to incredible opportunities early in my career that I don’t know if I would have found elsewhere. Our motivation for moving was primarily for personal reasons, but we asked ourselves many questions before pulling the trigger: > Can we find fulfillment and growth in our careers beyond the first role when we land? > What will our financial picture look like if we made $X? And spent $Y? > What does life outside of work look like? With family? And friends? > Is {healthcare|rental market|the TTC} really as bad as people say it is? For Canadian tech expats who are asking themselves similar questions especially in light of the recent announcements related to H1B, I’m happy to share my experience navigating the change and maybe some tips I learned from the whole process to make it less daunting 🇨🇦 #H1BNews #TorontoTech #LifeInToronto #Canada #CanadiansAbroad #H1B #TechToronto #Engineering

192

9 Comments

I’ve spent 10+ years building production systems across companies like Myntra, Curefit, and Atlassian. For most of that time, my value was clear: • design scalable systems • ship reliably • mentor teams • make trade-offs under real constraints Then AI happened. And if you’re a tenured engineer, you’ve probably felt one of these: 👉 “This feels like a grad-student game.” 👉 “I don’t want to start from scratch again.” 👉 “Isn’t this just prompt engineering + hype?” I felt all of that. But here’s what clicked for me 👇 AI doesn’t reduce the value of senior engineers. It rewards them—if we pivot the right way. Most real-world AI problems today are not about inventing new models. They are about: • turning fuzzy problems into clear interfaces • integrating AI into messy, legacy systems • evaluating correctness when answers are probabilistic • designing guardrails, fallbacks, observability • making AI useful at scale, not in demos That’s not beginner work. That’s applied AI engineering. And this is where experience compounds: • Knowing where AI should not be used is as important as knowing where it should • Understanding failure modes matters more than model accuracy • Product intuition + systems thinking beats raw ML theory most days The pivot doesn’t look like: ❌ “Become an ML researcher overnight” It looks like: ✅ “Use AI as a new primitive in systems you already know how to build” Senior engineers who win in the AI era will: • treat LLMs like unreliable distributed systems • design for cost, latency, safety, and user trust • bridge product, infra, and intelligence • ask better questions—not just better prompts We’ve reinvented ourselves before: monoliths → microservices servers → cloud manual ops → DevOps AI is just the next shift. The uncomfortable part? Letting go of being the expert again. The exciting part? The leverage is unlike anything we’ve seen. If you’re a tenured engineer navigating this transition: • What excites you about AI? • What scares you? • What are you actively learning—or resisting? Curious to hear how others are thinking about this 👇 #Engineering #AppliedAI #SoftwareEngineering #CareerGrowth #TechLeadership

25

Self-code review checklist I wish I had when I was a Jr. Engineer at Amazon: ○ Before I even run the code: – Do I fully understand the input, output, and constraints? –Did I rephrase the problem in my own words? –Did I list edge cases before touching the keyboard? –Did I pick the simplest working solution first? –Am I optimizing only after correctness is proven? ○ While writing the logic: –Are my variable and function names clear without comments? –Is each function solving only one responsibility? –Did I remove hard coded magic numbers? –Are my conditions readable at first glance? –Am I avoiding deep nesting where early returns work better? ○ After the first successful run: –Did I test empty input, single element, and max constraints? –Did I manually test worst case scenarios? –Did I validate both success and failure paths? –Did I remove all commented dead code? –Did I remove all temporary debug logs? ○ Performance and safety checks: –Is any loop unnecessarily nested? –Am I recalculating something that can be cached? –Is there any chance of null pointer or out of bounds access? –Is any blocking call happening on the main thread? –Are all error cases actually handled? ○ Readability and maintainability: –Can another engineer understand this without my explanation? –Are functions short and intention revealing? –Does the file structure feel obvious? –Are my comments explaining why, not just what? –Did I reduce complexity or just rename it? ○ Production readiness: –Are errors logged with enough context to debug later? –Are retries and fallbacks handled where needed? –Are timeouts set for all external calls? –Am I leaking any sensitive data in logs? –Would this survive real world traffic without babysitting? ○ Final gut check: –Would I confidently approve this in a real PR? –Would I be okay debugging this at 2 AM? –Would I understand this code after 6 months? –Would a senior engineer push back on this? –Did I actually improve the code or just rearrange it?

246

9 Comments

Sam Altman just dropped a deadline for the entire industry: "By 2028, data centers will hold more intelligence than ALL human brains combined." This isn't just a headline. This is a countdown. At the India AI Impact Summit yesterday, Altman made it clear: AI compute has grown 4.5x annually since 2012. The math checks out. We are approaching a point where the world’s collective cognitive power will be dwarfed by silicon. Most developers are preparing for the wrong thing. Here is what this actually means for your career: The value is shifting faster than your GitHub streak: From writing code → to understanding complex systems. From knowing syntax → to knowing exactly WHAT to build. From solo output → to orchestrating autonomous AI agents. The developers who thrive in 2028 won't be the ones coding more. They’ll be the ones thinking better. 🧠 The Cynic’s Reality Check: Let’s be honest: if you think buying a "Prompt Engineering" course is your golden ticket, you’re mistaken. "Intelligence in data centers" isn't a magic button for easy money. It’s a hyper-competitive landscape where only those who can actually control that power survive. The market for "just coders" won't die because of AI—it will die because the barrier to entry for mediocrity is hitting zero. Your 3-Year Strategy (The window is closing): You have roughly 36 months. That’s enough time to pivot if you start today: • Master Expert Orchestration: Move beyond simple chats. Build and test real-world AI agents. • Develop Judgment: AI can generate 1,000 solutions, but it can’t tell you which one won’t tank your business logic in six months. • Prioritize Alignment: The winners will be those who put safety and oversight first. Superintelligence won't kill developers. It will kill bad developers and unaligned agents. This is the best—and most dangerous—time to level up. Ever. Altman’s 2028 prediction: Is it the most exciting or the most terrifying thing you’ve heard this year? Drop your thoughts below. Let’s discuss. 👇 #AI #SoftwareEngineering #GenerativeAI #SamAltman

5

4 Comments

India could lose $140 billion annually if Trump’s proposed H-1B visa fee goes through. A suggested $100,000 fee on H-1B visas would effectively dismantle the program, which has been a lifeline for hundreds of thousands of high-skilled professionals—especially from India. 1. India is the world’s largest recipient of US remittances, receiving about $140 billion annually. These inflows make up nearly 30% of India’s tax revenues, supporting economic growth and currency stability. 2. About 71% of all H-1B visa holders are Indians. In 2025, ~399,395 H-1B visas were approved, of which nearly 285,000 went to Indians—primarily in the tech sector. This means nearly 300,000 Indian professionals in the US could be directly impacted. 💡 The consequences? A massive hit not just to Indian families and the Indian economy, but also to US companies that rely on top global tech talent to stay competitive. More Indian professionals may start shifting their focus toward Gulf countries like the UAE and Saudi Arabia, where tech ecosystems are rapidly expanding and policies are becoming more talent-friendly. #H1B #USIndia #TechTalent #Immigration #Economy #FutureOfWork #UAE #SaudiArabia

18

1 Comment

How I Became a Senior Software Engineer at Intuit Less Than 2 Years After College. I was never the “best” programmer in any room I walked into. Before college, I didn’t even know what coding was. And throughout college, I often struggled in my CS classes. But the beauty of working in tech is that everything keeps evolving — and if you keep learning, you’ll never be defined by what you haven’t learned yet. Despite my mediocre start as a developer, I was fortunate to become a Senior Software Engineer at Intuit within 2 years of graduating. Here are a few lessons and habits that helped me get there: ✅ Give your best to the current project While my classmates were stressing over full-time job applications during senior-year summer, I focused on delivering my internship project at Intuit to the best of my ability. That paid off: I got a return offer — and even bypassed the entry-level role to join as a Software Engineer II. ✅ Don’t just say “I’m blocked” and wait When I hit a roadblock, I didn’t sit and wait for help. I reached out, asked questions, and sought solutions. The engineers I admired didn’t just say they were blocked; they explained how they were unblocking themselves and moved forward. ✅ Keep challenging yourself I took on tasks that scared me or felt beyond my skillset. Instead of saying “no,” I figured out how to learn and deliver. That versatility and willingness to step up didn’t go unnoticed. ✅ Be responsive and helpful I made it a point to respond quickly to customer and teammate questions, spent time debugging with others, and helped on internal channels. This sharpened my skills, deepened my knowledge — and built a circle of people who were just as willing to help me when I needed it. ✅ Let your actions speak I didn’t ask my manager for a promotion. I focused on delivering high-quality work, owned up to my gaps, and showed how I was improving. I was lucky to have mentors and managers who noticed and guided me further. ✅ Be a good teammate At work, you’re not there to prove yourself — you’re there to solve problems together. Being resourceful, helpful, and collaborative goes a long way. On a personal note — I recently moved to #Singapore in June and am excited to build the next chapter of my career here. I’m currently exploring full-time opportunities as a Software Engineer or Machine Learning Engineer. If you happen to know of any openings or could connect me with someone in your network, I’d be so grateful — feel free to reach out anytime! #SoftwareEngineering #TechCareers #MachineLearning

104

25 Comments

Mark Zuckerberg just made a $1 billion dollar job offer to an engineer named Andrew Tulloch and was rejected. So who is Andrew? Andrew is a brilliant engineer who's worked at Meta before. After giving the company 11+ years of his career, he became a Distinguished Engineer [E9]. This is an insane accomplishment already. <100 people have ever become Meta E9, and the average E9 makes $2,500,000+ per year. However, Andrew is now at Thinking Machines, a leading AI startup founded by the ex-CTO of OpenAI Mira Murati, valued at $12 billion. Like many, Andrew has realized that if you genuinely want to change the world, you can't work for FAANG and its bloated bureaucracies and politics. Meta is an amazingly well-run company (just look at the stock price), but it's undeniable that its innovation muscle is long gone. Over the past 10+ years, most of its "new" ideas have been acquisitions or copy-pastes of competitors. Meta AI is mostly associated with slop. If you want to build the future of AI, the path is obvious - Work for a startup. You can get personally matched up with opportunities at top AI companies like OpenAI and xAI here: https://lnkd.in/g4Prv7jp #techcareergrowth #softwareengineering #growthtips #startups #faang #ai #meta

468

28 Comments

💸 “Congratulations! Your H-1B visa now costs the same as a Tesla Model S. Except you don’t get wheels… you just get to stand in line at USCIS.” That’s not satire. That’s reality. The Trump administration has raised the H-1B visa filing fee from around $2,000–$5,000 to a jaw-dropping $100,000 per application. The justification? Protect American jobs. The reality? This policy is about to: • Crush startups that can’t throw six figures at paperwork. • Slow down innovation in areas like AI and biotech. • Hand top global talent to Canada, Europe, and everywhere else rolling out the red carpet. Tech giants are already in damage-control mode: • Amazon warned H-1B employees not to leave the U.S. without a “$100K re-entry ticket.” • Microsoft urged workers to stay put. • Google told overseas staff to rush back before the gate slammed shut. Economists are calling it “anti-growth.” Lawyers are calling it an illegal tax. Immigration advocates? They’re calling it what it is: a brain drain, by design. Here’s the irony: you can’t conjure up a domestic STEM workforce by pricing out global talent. You can’t build the future if you’re busy erecting toll booths at the border. This isn’t just about visas. It’s about whether the U.S. wants to lead the next wave of innovation — or watch it happen somewhere else. Sometimes, the most expensive mistake isn’t the one you pay for. It’s the one you lose. #H1B #Immigration #Tech #Innovation #Leadership

6

Why does an E3 level SWE at Meta make only ~190k/year, while an E8 level engineer makes over ~$2M/year, even though both engineers are ICs and spend the same time at work? I have spent the last 5 years at Meta as an IC. I joined with a little over 15 years of experience, and I’ve worked with many solid engineers in this time. Here is how I think about that compensation jump. 1. Same hours, completely different “unit of work” An E3’s unit of work is usually a task or a ticket. An E8’s unit of work is a multi year problem for the company. E3: “Implement this service, fix this bug, write this feature.” E8: “How do we cut infra cost by 20 percent across this product” or “How do we make this platform safe to scale to 10x users.” One person is paid to execute. The other is paid to decide what is worth executing in the first place. 2. Radius of impact E3 usually impacts a file, a service, maybe a small team. E8 shapes whole orgs and product lines. If an E3 ships something great, the impact is great but local. If an E8 ships the right platform, hundreds of engineers become faster and the company saves or earns millions every year. Comp tracks the area of the circle you influence. 3. Risk and downside protection At junior levels, mistakes are usually contained and reversible. At senior staff levels, a bad call can burn tens of millions or damage the brand. E8s are paid for judgment under ambiguity. They decide which bets the company should not make, which migrations can wait, which “shiny idea” is going to kill reliability. You pay more to people whose good judgment protects you from very expensive failures.  4. They scale themselves This can happen in a few ways. 1. Delegation with ownership They define the shape of the problem, then hand large pieces to other senior and mid level engineers while keeping the bar and direction clear. 2. Knowledge that travels They write RFCs, public comments, FAQs, wikis, internal posts. One answer helps hundreds of people who will face the same issue next quarter. 3. Tools over heroics Instead of unblocking people manually all day, they build tools, libraries or guardrails so others can unblock themselves. One well designed tool can save thousands of engineering hours every year. This is what “scaling yourself” actually looks like. The company pays for that multiplier.  5. Ownership of the “uncomfortable problems” Junior engineers usually work inside a well defined box. E8s take ownership between the boxes. They pick up problems that:  Span many teams and no one really “owns”  Require aligning leaders who disagree  Have product, infra, legal and security angles at the same time Most people avoid those because they are messy, political and slow. Very senior ICs lean into them. That is where a lot of value sits. Continued ↓

201

15 Comments

🚨Despite the new $100,000 fee per H-1B visa, Jensen Huang confirmed that NVIDIA will continue sponsoring visas & cover all costs. His reason? “Immigration is central to America’s ability to lead in technology and ideas.” It’s a bold move and a powerful message. Leadership isn’t about cutting costs. It’s about standing by what drives innovation. If you’re a professional aiming for roles in AI or advanced tech, this signals two things: ➝ Top companies will always pay for the best talent, no matter the policy climate. ➝ Your value grows when your skills travel well, when you can adapt, collaborate and build across borders. For founders and leaders, it’s a clear reminder: Don’t let policies decide how you hire or grow. Let your vision lead the way. ➡️Take for students: Build expertise that transcends location. The world’s best companies look for capability, not just geography. 1️⃣Master one domain deeply, go beyond your syllabus. 2️⃣Build real projects and show your work online. 3️⃣Collaborate globally through hackathons or open-source. 4️⃣Focus on solving real problems, not chasing buzzwords. 5️⃣Communicate and teach what you learn, it builds clarity and credibility. Remember: Your strong skills are your passport. They can open doors anywhere in the world.

33

3 Comments

India needs to manufacture its own RAM. Not assemble. Not import and brand. Manufacture. Why this matters: - RAM is a core dependency for every phone, laptop, server, and data center India is almost entirely import-dependent for memory chips - Any supply disruption directly impacts cost, availability, and scale - AI, cloud, and compute growth will only amplify this risk This isn’t about nationalism. It’s about control over critical digital infrastructure. You can’t build a serious digital economy on borrowed memory. P.S. Yes, it is possible to manufacture RAM in India — but not overnight. It requires: • Massive capital investment • Advanced fabrication plants (fabs) • Long-term policy consistency • Global technology partnerships It’s hard, expensive, and slow — but strategic, and eventually unavoidable. Semiconductors built the last decade. Memory will define the next one.

4

2 Comments

Einstein didn’t use '8 years' of 'test-time compute.' This is a category error. While the test-time overhang is real, the scaling at inference has brutal diminishing returns (more so than pretraining). Nobody's torching 8 years of compute praying for Physics 2.0. This is science fiction.

203

91 Comments

If you are an SF engineer at $150K, you are likely Level I. Your H‑1B chances just dropped by 48%. The new H‑1B Cap Weighted Selection Final Rule is here. Random lottery is gone. Wage‑weighted selection is in. Data. Under the old system, everyone sat at about 35% odds. 118K out of 336K registrations in FY26. No advantage for higher wage levels. The new rule rewrites those odds: → Level I: 15.29% → Level II: 30.58% → Level III: 45.87% → Level IV: 61.16% For Level I, odds drop nearly in half. For Level IV, odds more than double. A $150K SF engineer is often Level I on OES. → Early career people, Lower salary but important roles, they get hit the hardest here. There is more. Registrations must now include: → OEWS wage level → SOC code → Work location Later the petition must prove that wage level was correct on the day of registration. USCIS can deny, revoke, or block amendments if it looks like you changed terms to game the lottery. The logic from the government is clear. → Higher pay, higher odds. → Your economic value now shapes your chance to stay. If you are an employer, start planning before the next window. If you are a worker, learn your wage level and study other paths, not only H‑1B. I will keep sharing data as more details come out. Hi! My name is Sumit Gupta, I share data, AI, immigration updates and make you a smarter immigrant!

33

2 Comments

🚀 India just quietly triggered one of the biggest tailwinds for AI Product Careers - and many PMs might not have noticed yet. In the last two weeks, three announcements landed that will directly accelerate AI product building in India: 1) ₹1 Lakh Crore DeepTech RDI Fund (Govt. of India) A national ₹1,00,000 crore innovation fund will support AI, robotics, quantum & advanced tech - including risk capital at early stages. This means a new wave of AI-first products will get funded, and they’ll need PMs who can convert technology into usable customer value. 2) Karnataka Startup Policy 2025–2030 (₹518 Cr for 25,000 Startups) The policy aims to support 25,000 tech startups (10,000 outside Bengaluru) with priority on AI, blockchain, space tech, and quantum. This signals more 0→1 product roles across India -not just in Bangalore. 3) Nvidia joins India Deep Tech Alliance Backers now include Nvidia + venture partners to catalyse Indian deep-tech ecosystems with not just capital but also computational and GTM support. 💡 What this means: India is about to over-fund AI ideas and under-supply PMs who can turn those ideas into real products. If you’re early in your career, this is the moment to: pick a problem space you care about understand where AI creates 10× value over status quo build & ship velocity instead of perfection The next generation of strong PMs won’t be defined by: “how well they execute Jira” but “how well they convert tech → customer value → revenue” That window just opened. #AIProductManagement #Productmanagementcareers #AI #DeepTech

42

3 Comments

India has ~20% of the world's chip designers. We have almost zero global chip companies. Yet. That gap isn't a talent problem. It's a courage problem and I want to explain exactly what I mean by that. For decades, the best chip engineering minds in India have built world-class products. For global companies. From India. Intel. Nvidia. Qualcomm. AMD and more. The silicon inside your phone, your laptop, your car - a meaningful slice of it was designed here. By Indians. But here's what didn't happen: we were never asked what the product should be. Only how to build it. And over time, we stopped asking that question ourselves. The courage I'm talking about isn't bravado. It's the specific courage to conceptualise a product from scratch. To walk into a distributor in Ritchie Street and ask what's selling. To build a supply chain for something nobody has ordered yet. To fail at version one and fund version two. That muscle was never built. Not because the talent wasn't there, but because the roles that would have built it simply didn't exist in India. G S Madhusudan, co-founder and CEO of InCore Semiconductors, one of India's first processor IP companies said it precisely on Chai & Chips Podcast: "The courage to conceptualize a product and push through does not come from raw courage. It also comes from experience." We were never given the chance to build that experience. That's the gap. It's changing. Slowly. Unevenly. But changing. The founders building chip startups in India today aren't more talented than the generation before them. They're just the first generation for whom product ideation was a real option, not a foreign concept. That's the shift worth watching. Not the fab announcements. Not the PLI numbers. The moment India stops executing other people's ideas and starts betting on its own. We're at the start of that moment. Not the end. Chai & Chips is India's only podcast about its semiconductor and AI journey. You can find us on YouTube, Spotify and Apple Podcasts. YT channel link in comments.

39

4 Comments

Do not over-rely on LLMs to write your code. Seriously. Relying too much on instant answers is the quickest way to forget how to think critically and solve problems on your own. Write the code yourself. Use LLMs to help review, debug or look up unfamiliar APIs. Not as a crutch for every task. Read the docs. Read books. Build things. Break things. Master the fundamentals, there are no real shortcuts. Happy coding 👨‍💻 #softwareengineering #llm

2,464

108 Comments