Nitish Upreti - San Francisco Bay Area | Professional Profile | LinkedIn

Soundarya Balasubramani

Curious Maverick • 130K followers

We're leaving India to work at FAANG, while FAANGs are investing tens of billions into infrastructure in India. Context: Google just invested $10B in India to build its AI infrastructure, and they are not alone. Microsoft added $17.5B and Amazon threw in $35B for the same reason. 📍 Some key highlights: → Building a data center in India costs $7 per watt. → A mid-level AI engineer in India makes $30K-$55K. → The same person in the U.S. makes around $180K-$250K. → That's 70% cheaper for the same work. 📍 Here’s more highlights: → India graduates 850,000+ engineers every year. → The U.S. graduates ~140,000. → Even Satya Nadella thinks India will have more AI talent than the US by 2027. → And AI models don't care where they're trained. Google, Microsoft and Amazon aren't just building their satellite offices here. They're also investing in building AI infrastructure in India because it makes better business sense. Between 2021-2023, nearly 1,000 international companies set up in India. All of this could mean: The career gap that used to justify any cost to reach the U.S. is narrowing fast. — 🌿 Re-post if this resonated. 👉 I share weekly stories of impactful people, companies, and art. Join 34,000+ at Curious Maverick

Arvind Kumar

Axis Bank • 4K followers

💣 GitHub CEO: AI Will Write 90% of Code by 2030 Max – Adapt Now or Be Left Behind. For developers across India and the world, the message is loud and clear: Reinvent or risk irrelevance. AI is no longer “the future”. It’s the present & we are entering hard tech era . From Bengaluru to San Francisco, developers are already witnessing the shift: AI tools are writing code, debugging, generating test cases, and even suggesting architectural decisions. But instead of job losses, something unexpected is happening — 💡 Developers are becoming more ambitious, creative, and strategic. A recent field study by GitHub CEO Thomas Dohmke reveals a fascinating transformation underway: 🔄 4 Stages of AI Evolution for Developers: 🔹Skeptic: Tries AI tools once or twice, calls them “cool but buggy.” 🔹Explorer: Starts using AI for small tasks — snippets, debugging. 🔹Collaborator: Begins co-creating with AI, working across files, tools, and models. 🔹Strategist: Designs systems, orchestrates agents, reviews AI output — and focuses on outcomes, not just code. One participant said it best: 🧠 “We’re becoming less code producers, more code enablers. My next title might be Creative Director of Code.” 🚨 Key Insights From the Study: 🔹 AI is on track to write 90% of code in the next 2–5 years. 🔹 Developers who embrace AI feel more optimistic and confident about their careers. 🔹 The focus is shifting from writing code to designing systems, prompting AI agents, and validating results. 🔹 AI fluency is the new must-have skill – not just how to code, but how to collaborate with AI. 🎓 Implications for Education (India, take note 👇): Teaching just syntax or theory won’t cut it anymore. CS students and aspiring engineers must now learn to: ✅ Prompt and guide AI effectively ✅ Debug and critique AI-generated code ✅ Think in systems, not just scripts ✅ Collaborate with AI tools, not avoid them Assignments must shift from: “Write a loop” to “Here’s what AI wrote — what’s wrong with it?” or “Improve this spec so AI can build what you intended.” 📈 Job Outlook? Surprisingly Strong. While traditional roles may evolve, the U.S. Bureau of Labor Statistics still projects an 18% growth in software developer jobs — nearly 5x faster than the average occupation. The same trend echoes globally, including in India’s booming tech sector. But the definition of a “developer” is changing. The most successful ones are: 🔧 Orchestrating AI agents 🧠 Applying deeper product thinking 📦 Building systems, not just code files ✅ So whether you're in Bengaluru, San Francisco, or Singapore — here's the reality: AI isn’t here to replace you. It’s here to work with you — if you know how to use it. #AI #SoftwareEngineering #GitHub #IndiaTech #FutureOfWork #DeveloperLife #AIRevolution #PromptEngineering #TechCareers #EdTech #CreativeDirectorOfCode #India #IT #hardtech

Nitin Sharma

Self-employed • 990 followers

Oracle is going to lay off employees in the US and India. Not because the company is failing but because it’s betting everything on AI. They’re cutting roles in their cloud division to free up billions for Project Stargate — a $500 billion AI infrastructure partnership with OpenAI and SoftBank. In plain English: Oracle is building the “factories” where the next generation of AI will be trained and run. Well, this is becoming the playbook for big tech: → Cut human costs in areas where AI will soon outperform. → Reallocate capital into AI infrastructure, models, and platforms. → Capture the future market before competitors do. The reality? • AI is moving faster than any past tech revolution. • It doesn’t just replace jobs, it eliminates entire categories of work. • As AI capabilities explode, companies will grow bigger, more efficient… with far fewer people. You see, we’re heading into a future where: → A handful of companies control the AI infrastructure. → Most routine jobs vanish. → The value of your skills depends entirely on how “AI-proof” they are. The harsh truth? We’re heading into a future where growth and profits will come from machines, not manpower. The winners will be companies that adapt fastest, and individuals who learn to work with AI, not compete against it. 𝗪𝗮𝗻𝘁 𝘁𝗼 𝗹𝗲𝗮𝗿𝗻 𝗺𝗼𝗿𝗲 𝗮𝗯𝗼𝘂𝘁 𝘂𝘀𝗶𝗻𝗴 𝗔𝗜 𝗽𝗿𝗮𝗰𝘁𝗶𝗰𝗮𝗹𝗹𝘆? 𝗦𝘂𝗯𝘀𝗰𝗿𝗶𝗯𝗲 𝘁𝗼 𝗺𝘆 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿: https://lnkd.in/dcx8YBYP #AI #artificialintelligence #Startups #DeepMind #OpenAI #jobs #technology #productivity #oracle

Simardeep S.

Department of National… • 4K followers

Pinterest 𝗝𝘂𝘀𝘁 𝗦𝗵𝗼𝘄𝗲𝗱 𝗛𝗼𝘄 𝘁𝗼 𝗔𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗠𝗶𝗴𝗿𝗮𝘁𝗲 𝗳𝗿𝗼𝗺 𝗛𝗮𝗱𝗼𝗼𝗽 𝘁𝗼 𝗞𝘂𝗯𝗲𝗿𝗻𝗲𝘁𝗲𝘀 (𝗔𝗻𝗱 𝗡𝗼𝘁 𝗠𝗲𝘀𝘀 𝗜𝘁 𝗨𝗽) Pinterest dropped a detailed write-up on how they moved their entire data processing platform from Hadoop to Kubernetes. 70% migrated already, rest coming by year-end. Instead of just moving Spark jobs to K8s and hoping for the best, they built "Moka" - a proper platform with all the pieces in place: job submission, resource management, monitoring, the works. They're running 𝗔𝗥𝗠 𝗮𝗻𝗱 𝘅86 𝘀𝗶𝗱𝗲 𝗯𝘆 𝘀𝗶𝗱𝗲. Not everything works great on ARM, but when it does, the cost savings are real. They're picking the right instance type per workload instead of going all-in on one architecture. 𝗔𝗽𝗮𝗰𝗵𝗲 𝗖𝗲𝗹𝗲𝗯𝗼𝗿𝗻 for shuffle replaced Hadoop's external shuffle service and delivered 5% better performance across the board. 𝗬𝘂𝗻𝗶𝗞𝗼𝗿𝗻 𝘀𝗰𝗵𝗲𝗱𝘂𝗹𝗲𝗿 brought all the queue management and resource controls from YARN over to Kubernetes - critical for keeping SLAs during the migration. They built a "dry run" system that automatically runs every production job on both the old Hadoop cluster AND the new K8s platform, then compares the results. Data sizes, checksums, everything. Only migrate jobs that pass validation. This took years and involved teams across the entire engineering org. Moving from Java 8 to 11, recompiling libraries for ARM, making sure containers work exactly like the old Hadoop setup - there's no shortcuts at this scale. But they pulled it off without breaking production, which honestly is the real achievement here. Worth reading the article by Soam Acharya, Rainie Li: https://lnkd.in/eYN_gmNX #BigData #Kubernetes #Spark #DataEngineering