Show Notes
Hari Prasad Renganathan got rejected from the H1B lottery three times. Most people in that situation spiral. They scramble for a second master's, beg their company for an internal transfer, or start mentally preparing to go back home. Hari texted 10 YC startups, got callbacks from four of them, landed an offer as a Senior AI Engineer, and moved to London. The UK work visa took three weeks. No lottery. But here's the thing — that's just the final chapter of a pattern that's been repeating his entire career. Every time Hari hit a wall, he did the thing nobody else was willing to do. He applied to Columbia with a mechanical engineering degree from Tamil Nadu. No CS background. His GRE score was so average he deliberately withheld it from his application. He wrote an SOP that broke every template the admissions consultants swear by. He got in. After Columbia, he couldn't land an interview for four months. When a senior data scientist on his team quit, Hari walked up to the decision maker and said — and I'm quoting him here — "I'll be much cheaper than that guy." He got the full-time offer. Then he launched My Real Product, a community workshop where he helped data scientists and AI engineers stop building cookie-cutter Kaggle projects and start building actual products with real users. He put up a LinkedIn post expecting 20 people. 470 showed up. We also get into the AI conversation. Hari has a take on the whole "AI won't replace you, but people using AI will" thing. He disagrees. So do I, actually. He breaks it down into four layers of AI replacement and where you need to be if you want to be the last person standing. Whether you agree or not, it's the most specific framework I've heard anyone lay out on this topic. This conversation is for you if you're an international student navigating the US visa system, if you're trying to break into data science or AI without a traditional CS background, if you've been following every template and wondering why nothing is working, or if you just want to hear what happens when someone refuses to take no for an answer at every single turn. Timestamps, links, and ways to connect with Hari are all below. If this episode helped you think differently about even one thing, share it with somebody who needs to hear it.
Site: readysetdopodcast.com Hari: https://www.linkedin.com/in/hariprasad20/
Timestamps: 00:00 Introduction & Connection 03:23 AI, Personal Branding, and LinkedIn Presence 05:29 From Mechanical Engineering to Data Science 10:11 Navigating the Columbia Application Process 14:56 Life at Columbia University 19:43 Transitioning from Student to Professional 23:26 Entrepreneurship and Building MyRealProduct 26:26 Problem Solving with Data Science 29:38 Building a Data Science Community 33:12 Rejection from H-1B Lottery and Exploring Career Paths
45:43 The Future of AI and Its Impact
More from Ready Set Do
Thirteen hundred applications. Thirteen callbacks. At one point, that was Bharath Vemula's job search math. Today he's a Meta engineer working on the company's model training stack. (The twist? He never applied to Meta. Recruiters found him on LinkedIn.) So what was on that profile? We pull it apart, along with the LinkedIn habit that got him headhunted. If you've ever wondered whether anyone reads your headline, this one's for you. His path runs through four years as a software engineer at Goldman Sachs in India, then Verily, the Alphabet moonshot company, where he worked on privacy and data governance. Then Meta. We go deep on the Meta interview process as it stands now. Two DSA rounds, a system design interview, a behavioral round, plus the AI coding round that's been added since. Is LeetCode dead in 2026? He says no. His reasoning is worth hearing even if you hate grinding problems (and yes, most of us do). Then we rewind to 2022. Bharath came to the US for his MS at Purdue with the whole plan already mapped out before he left India. He turned down Columbia over the fee and picked Purdue for the assistantship odds. A research role in a cyber safety lab funded most of his master's, and he finished in eighteen months. His one rule? Walk out with zero debt. Study abroad folks, take notes. This is the international student playbook most people wish they had on day one. We also get into where big tech jobs are opening up next. His bet is ML systems engineering and AI infrastructure. He thinks the compute wasted across big tech data centers is one of the biggest openings for engineers right now. In the back half, we get EB-2 NIW explained in plain terms. How a green card NIW petition skips PERM, what USCIS looks for, and why he started it as protection against tech layoffs before it became the whole plan. For any Indian engineer in the USA, it's an H-1B alternative worth understanding. Plus his side research on AI agent security, how agents change the economics of social engineering attacks, and how to build a research profile without a university behind you. We close on the big question. Is AI a bubble, and will it take your job? His answer is more specific than yes or no. If you're deep in a software engineer job search right now, watch the 1,300 applications chapter twice.
Updated to 3,750 characters total, chapters included, so you have about 250 to spare. Akash prepared full time for India's UPSC civil services exam for two years. Eight to nine hours a day, two attempts. He didn't make it. Then he gave himself three days to decide on a third try, and barely slept through any of them. He walked away. (Spoiler: what came next was far messier than "UPSC aspirant becomes software engineer in the US.") Why is UPSC such a trap? Akash calls it a black hole for India's youthful years. The exam is binary. You clear it, or you leave with zero. He shares the four questions to answer before another UPSC attempt. He also names the sunk cost trap that keeps aspirants coming back long after they should stop. At 24, he held a metallurgy degree from IIT BHU and zero coding background. He weighed CAT and GATE against the GRE, said no to an MBA, and chose an MS in CS in the USA. How do you build a CS profile from nothing? He co-wrote a machine learning paper with a professor back at his IIT. He also covers how to explain a two-year gap on applications. Then he applied to 17 US universities. One admit. Would you guess the next part from Instagram? Fear of failing a course on an F1 visa. No return offer. A layoff. $100K in student loans. (The Grand Canyon photos leave that out.) For international students job hunting in the USA, he breaks down 20 to 30 applications a day and what a real referral looks like. If you're sitting on a decision you've already spent years paying for, this one is for you. Subscribe on YouTube and Spotify, or visit readysetdopodcast.com.
A week before we recorded this, my nephew asked why I'd started posting on LinkedIn. He'd just finished college and started his first job, so I gave him the textbook answer about networks and opportunities. (I watched it bounce right off him.) Fair enough. If you're a new grad deep in a graduate job search, "build your network" sounds like homework with no due date. So I brought in someone who could answer him properly. Shoaib Ahmed is a LinkedIn ghostwriter who has written over 20,000 posts. His first job out of university came through a LinkedIn DM, and he later used the platform to build a million-pound business. (So yes, he's earned an opinion.) Why does LinkedIn personal branding matter more now that AI writes half the feed? Shoaib's answer surprised me. The flood of AI slop is your opening. When every post sounds the same, a real voice gets noticed fast. For personal branding beginners who don't feel like experts yet, he covers what to post anyway. And this isn't a founder-only game. Employees build a personal brand too, and he explains how. Then we get into the business side. Clients say they want one thing and actually want another. Impressions are a vanity number. You need reach and conversion, and one without the other stalls out. His LinkedIn profile audit covers the banner, the offer, your ICP, and the proof. Sell the transformation instead of the service. Make your ideal buyer feel seen. Let testimonials and case studies do the bragging. If your profile still reads like a résumé nobody finishes, jump to 00:37:13. If you're a student figuring out how to use LinkedIn, or you want to get hired on LinkedIn without sounding like a robot, this is the career advice I wish I'd handed my nephew. His closing line stuck with me: your LinkedIn profile is not for you. 🎙️
Shrey Shah never applied to Microsoft. Microsoft executives watched him give a talk on AI coding workflows, then reached out. Meta came knocking not long after, asking for something similar. Here's the part that stings a little: that wasn't luck. It traces back to a decision he made years earlier, when senior engineers around him said going all in on AI-assisted coding would hurt his career. He ignored them. He was building with agents before agent jobs existed, and using Cursor before most engineers could spell it. In this episode of Ready Set Do, Shrey — now Senior Software Engineer, AI at Microsoft — walks through how inbound offers actually happen in tech, and what he did to make himself findable. We get into the interview loop too. Four rounds, two technical, zero LeetCode. One problem was so big he had to keep solving it out loud while talking to his interviewer. He explains why that format tests something the standard algorithm grind never touches. Then the harder conversation starts. Shrey doesn't believe in resumes anymore. His argument: every resume is now tailored by the same AI tools, every application lands in the same ATS pile, and the traditional job application is finished. (If you're applying into silence right now, you already suspect he's right.) So what replaces it? Visibility. Posting what you learn. Forming your own point of view on what you read. Building a public track record long before you need one. He gets specific about what to build, what to post on LinkedIn as a software engineer, and how he never runs out of things to say. He also breaks down how meetup and conference speaking invitations actually arrive. Then the technical gut-check. Most engineers using AI tools can't explain the difference between a skill, a sub agent, a hook, a rule, and MCP — or when to reach for which. Shrey can, and he lays out the architecture in plain language. We close on his 2028 prediction. He doesn't think AI takes your job. He thinks roles merge, everyone carries more scope, and the new jobs created won't line up with the ones that vanish. Worth your 45 minutes if the job market has gone quiet on you. 🎧
Thirteen years at Google. Trust and Safety in Hyderabad first, then the Bay Area, then most of the back half inside YouTube — finance, then analytics, then managing YouTube inventory and monetization. Then Aishwarya Raghavan quit. No job lined up. Two months later a friend referred her to a company she actually wanted. She didn't get it. That rejection landed right as the layoff headlines were peaking, and that's where this conversation starts. Most advice about the AI job market in 2026 is built to make you feel behind. Aishwarya's take goes the other way. She says a lot of postings that say "AI skills required" mean something far more basic than you'd guess. The people who never start are usually the ones who jumped straight to AI agents, decided it was too hard, and quit before level one. (If that's you, no judgment. It was most of us.) So she walks through four levels anyone can climb. Level one is small enough to feel silly: getting ChatGPT to stop sounding like a stranger. Custom GPTs come later. Agent orchestration comes after that. Is that really enough to get hired in AI? Her answer is more specific than yes. We also get into why she couldn't just move into AI work internally at Google, even with the badge and the tenure. How to read a job description and tell whether a company has actually thought about AI or is checking a box. And the interview skill she rates above every technical signal. Fair warning — it has more to do with storytelling than with reciting the STAR method at a hiring manager. Then there's the networking habit she regrets not building while she still had the Google email. Nobody tells you that your network has an expiration date. She's now a principal product manager at Arrivia, wiring AI into a travel business full of legacy systems. Different problem entirely from YouTube creator monetization, and she's candid about what transferred and what didn't. If you're a student, or a few years into your career and quietly worried the market already moved past you, start here. Chapters below. If this helped, subscribe on YouTube and Spotify. 🎧 00:00 - Aishwarya Raghavan on Leaving Google After 13 Years 04:12 - How AI Is Changing Big Tech and Product Management 07:14 - Product Manager Job Search Advice in a Tough Tech Market 10:09 - Product Interview Storytelling: How to Stand Out as a PM 13:24 - Why Product, Engineering, and Strategy Roles Are Blending 16:10 - Best AI Tools and Workflows for Product Managers 19:08 - AI Agents and Agent Orchestration: The Next PM Skill 22:03 - Job Search Lessons After Leaving Google 25:17 - Travel Tech and AI: Modernizing Legacy Product Workflows 30:45 - YouTube Monetization, Creator Tools, and AI Product Strategy 35:59 - Networking Without an Agenda for Big Tech Career Growth 40:23 - Final Advice for PMs Building Rare AI Skills