<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Divyanshu Rajput on Medium]]></title>
        <description><![CDATA[Stories by Divyanshu Rajput on Medium]]></description>
        <link>https://medium.com/@divyanshurajput709?source=rss-0942ae804bbe------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*QQ_3OKSY9b0J8WAZYSsJ6A.jpeg</url>
            <title>Stories by Divyanshu Rajput on Medium</title>
            <link>https://medium.com/@divyanshurajput709?source=rss-0942ae804bbe------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 13:59:01 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@divyanshurajput709/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[My Interview Experience: Turtlemint (SDE Intern, Onsite Pune)]]></title>
            <link>https://medium.com/@divyanshurajput709/my-interview-experience-turtlemint-sde-intern-onsite-pune-9a162aa5505e?source=rss-0942ae804bbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/9a162aa5505e</guid>
            <category><![CDATA[finance]]></category>
            <category><![CDATA[interview]]></category>
            <category><![CDATA[jobs]]></category>
            <category><![CDATA[turtlemint]]></category>
            <category><![CDATA[insurance]]></category>
            <dc:creator><![CDATA[Divyanshu Rajput]]></dc:creator>
            <pubDate>Thu, 17 Sep 2026 19:12:53 GMT</pubDate>
            <atom:updated>2026-09-18T04:29:25.596Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/738/1*SYAQ4l7n8ygs6KQhh1Wc5w.jpeg" /><figcaption>Turtlemint</figcaption></figure><h3>How it started</h3><p>There was no job opening. That’s the part worth leading with.</p><p>I reached out to a Principal Software Engineer at <strong>Turtlemint </strong>asking for a referral, even though I hadn’t seen any posted role — and somewhere along the way I learned that this is actually how they hire a fair bit of the time.</p><p>At that point, they genuinely had nothing open. But 2 months later, he messaged me out of the blue: he’d submitted my referral, and I’d be hearing from HR about next steps.</p><p>There was no portal, no application form, nothing. It was a direct referral from the person I’d actually be working under. (All of this happened before Turtlemint’s IPO.)</p><p>About a week later, HR called to say I’d been shortlisted.</p><p>The OA</p><p>Before the interview itself, there was an online assessment on HackerEarth: 2–3 easy-to-medium DSA problems. I solved all of them.</p><p>I can’t recall the exact problems now, but they were straightforward array and string questions — nothing fancy like DP or graphs.</p><h3><strong>The Interview</strong></h3><p>Here’s what made this one different: I was interviewed by the same person who referred me — the PE himself.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*obNVDO4mN70LJj9LmGwbNQ.png" /><figcaption>Interview Invite — Turtlemint</figcaption></figure><p>It started with my introduction and then went straight into the system design of AI agents. We worked through problems and designs together, and his focus was relentlessly on the reasoning:</p><p>What are you using? Why that? How does it work? What would a better solution look like? Why this approach and not that one, for now?</p><p>Every question built on the answer I’d just given. There was no jumping between unrelated topics — it was one continuous thread, which meant I couldn’t coast on a good answer or hide behind a bad one. It kept the whole conversation honest.</p><p>He was kind throughout, and genuinely clear in how he communicated — the kind of interviewer who wants to understand how you think, not catch you out.</p><h3>The Outcome</h3><p>I didn’t make it through, but the reasons were mostly circumstantial:</p><p>Business requirements had shifted on their end. They already had a lot of interns at that point and couldn’t take on more. My availability didn’t line up—my 6th semester exams fell right in the middle of the internship period, and Bareilly is a long way from Pune.</p><h3><strong>What I actually learned</strong></h3><p>This is the one that stuck with me, and it’s the reason I wanted to write this post at all.</p><p>Be real in an interview. Don’t frame your answers to sound impressive.</p><p>I caught myself doing exactly that during this interview — giving bookish answers, repeating things I’d absorbed from YouTube videos and other resources rather than saying what I actually thought. And when you’re being interviewed by someone genuinely experienced, that gets spotted immediately.</p><p><strong>Verdict</strong>: Rejected</p><p><strong>LinkedIn:</strong> <a href="https://linkedin.com/in/divyanshu-rajput8171">https://linkedin.com/in/divyanshu-rajput8171</a><br><strong>Portfolio:</strong> <a href="https://divyanshu-rajput.vercel.app/">https://divyanshu-rajput.vercel.app</a></p><p><strong>Fun Part:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S73vOnCxEjfk6F6yzONhMA.png" /><figcaption>Rejection email they sent</figcaption></figure><p>I got the rejection email for the SSE role, but I interviewed for the SDE Intern role, so after all the steps, I got to know that I was interviewed for the SSE role, not for the role I applied for, due to miscommunication by the hiring team.</p><p>What I appreciated most was that he didn’t call it out harshly—he just kept steering the conversation until I was giving him my own real thinking instead of someone else’s. That shift, mid-interview, taught me more than the outcome did.</p><p>If an experienced person is interviewing you, you will get caught trying to sound like someone you’re not. Save yourself the effort and just be honest about what you know, what you don’t, and how you’d actually approach it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9a162aa5505e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Interview Experience: Minerva (YC S25) — Contract Role]]></title>
            <link>https://medium.com/@divyanshurajput709/my-interview-experience-minerva-yc-s25-contract-role-d8c368777743?source=rss-0942ae804bbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/d8c368777743</guid>
            <category><![CDATA[ycombinator]]></category>
            <category><![CDATA[jobs]]></category>
            <category><![CDATA[meetings]]></category>
            <category><![CDATA[interview]]></category>
            <category><![CDATA[hiring]]></category>
            <dc:creator><![CDATA[Divyanshu Rajput]]></dc:creator>
            <pubDate>Wed, 16 Sep 2026 18:23:05 GMT</pubDate>
            <atom:updated>2026-09-18T04:30:03.502Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*3LJxMu7c7PAtmXr6L9gVkQ.png" /><figcaption>Minerva-YC25</figcaption></figure><p>This was my <strong>second interview with a YC-backed company</strong>, and the longest, most involved process I’ve been through so far — it stretched across multiple rounds, a take-home assessment, and even a short trial period working alongside the team.</p><h3>How it started</h3><p>Like a few of my other stories, this one started with a cold LinkedIn message to the founder.</p><p>And then I waited. For a long time. <strong>7–8 months later</strong>, I got a reply — he’d liked my profile and sent over a cal link to schedule an interview. A good reminder that cold outreach doesn’t always die in someone’s inbox; sometimes it just sits there until the timing lines up on their end.</p><h3>Round 1 — with the CTO/co-founder</h3><p>The round opened with my intro, and then the founder introduced himself properly — what Minerva is building, and what he’d been doing before it. He’d previously run a business in Canada, and after getting into YC, built Minerva.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vzmS7PXU_coGyPQrrFgwaw.png" /><figcaption>Round 1 — CTO Round</figcaption></figure><p>From there, he asked about my <strong>competitive programming journey</strong> — and specifically, why I’d stopped. My honest answer: I got into CP because it felt like a <em>sport</em>. Once AI usage became widespread in contests, that feeling went away. It stopped feeling like I was competing against other people, and once that was gone, so was my interest in it.</p><p>We then spent about 10 minutes on my projects, and by the end he said he was interested enough in my profile to move me forward with an assessment.</p><h3>The assessment</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*D0MZUUBp0jQQ6T2u2C-tfw.png" /><figcaption>Assessment — Minerva YC25</figcaption></figure><p>The task was to build an <strong>automation system</strong> for a specific problem they gave me. It wasn’t especially hard on its own — what made it interesting was the framing:</p><ul><li><strong>AI use was explicitly allowed</strong>, as much as I wanted</li><li>I had a <strong>full week</strong> to submit</li><li>Along with the solution, I had to record a <strong>video walking through and explaining it</strong></li></ul><p>That last part mattered more than I initially realized. When AI use is unrestricted, the code alone tells you less about the candidate — the explanation is where the actual thinking shows up.</p><h3>The trial</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jJbwv73NltGHEWHMA9F-Rw.png" /><figcaption>Trial — Minerva (YC 25)</figcaption></figure><p>After I submitted, the founder called to say they were interested in moving forward — but rather than jumping straight to another interview, they proposed a <strong>short trial working with the team</strong>. The goal was to see how well I could communicate and actually collaborate day-to-day, not just how I perform in an interview setting.</p><p><strong>First up: a senior software engineer</strong> at Minerva — Indian, from Delhi, and genuinely one of the most humble people I’ve interacted with in a process like this. He walked me through everything: the org, what they’re building, how they’re building it, and the real problems they’re currently facing. Nothing felt gatekept.</p><p>He then gave me a problem to solve — but not a coding problem. It was <strong>research-oriented</strong>: go investigate this, come back with findings. I did the research and submitted it back to him.</p><p><strong>Next: a call with the CEO/co-founder.</strong> I’d already gotten some context on him from the SSE, including a useful tip — listen to him fully before answering, since that would lead to better responses on my end. (The CTO being Indian meant communication was never an issue on that front either.)</p><p>The call started with intros, then moved into a <strong>LeetCode-style question that was fairly easy</strong>. I gave it a real attempt and discussed several approaches, but couldn’t land on a solution that was both correct and acceptable for the problem. After that, we talked more about the org generally.</p><h3>The outcome</h3><p>About a week into working within the org, I got the rejection.</p><p>The reason was straightforward: I was <strong>too early in my career</strong>. They needed people with more experience, and I am still in college. Worth noting, this was a <strong>full-time remote role</strong> — I’d seriously considered joining, since my college has no attendance requirement, so the logistics would have actually worked out.</p><p><strong>Verdict</strong>-Rejected</p><p><strong>LinkedIn:</strong> <a href="https://linkedin.com/in/divyanshu-rajput8171">https://linkedin.com/in/divyanshu-rajput8171</a><br><strong>Portfolio:</strong> <a href="https://divyanshu-rajput.vercel.app/">https://divyanshu-rajput.vercel.app</a></p><p><strong>What they are paying: 2K USD—5K USD based on the experience</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/738/1*2Qe9vbc0q5E_Fr3diwO1yw.png" /><figcaption>Y Combinator</figcaption></figure><h3>What I took away</h3><p>Even though it didn’t work out, this was one of the more valuable processes I’ve been part of. Beyond the interviews themselves, I got to talk with the founders about some of the <strong>internal details of how they got into YC</strong> — what actually works, what YC looks for when deciding to back a startup, and what matters in that process.</p><p>Getting a window into that, plus the chance to connect with people this thoughtful, made the whole thing worth it regardless of the outcome.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d8c368777743" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Interview Experience: Litmus (YC ‘26)]]></title>
            <link>https://medium.com/@divyanshurajput709/my-interview-experience-litmus-yc-26-bbffc63fc2ab?source=rss-0942ae804bbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/bbffc63fc2ab</guid>
            <category><![CDATA[assessment]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[ycombinator]]></category>
            <category><![CDATA[hiring]]></category>
            <category><![CDATA[interview]]></category>
            <dc:creator><![CDATA[Divyanshu Rajput]]></dc:creator>
            <pubDate>Tue, 15 Sep 2026 14:44:37 GMT</pubDate>
            <atom:updated>2026-09-15T14:44:37.960Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/150/1*1rj6jeSdwe3OO8aJAG7Q4g.png" /><figcaption>Litmus (YC 26)</figcaption></figure><p>This was my <strong>first-ever interview with a YC-backed company</strong>, and honestly one of the more memorable processes I’ve been through. Most interview stories I share follow a fairly predictable shape — resume, DSA, system design, feedback. This one didn’t, and that’s exactly what made it worth writing about.</p><h3>How it started</h3><p>I sent a LinkedIn message to the founder, requesting an internship — nothing fancy, just a cold, direct ask. Then… nothing, for a while.</p><p><strong>2–3 months later</strong>, out of nowhere, I got a reply. They weren’t hiring interns at that point, but they did have an opening for a <strong>contract role</strong>. If I was interested, the next step was simple: send my portfolio and projects to the email they shared. No formal application portal, no multi-page form — just an email.</p><p>I sent it over, and 2–3 days later got word that I’d been shortlisted for the next round. Fast turnaround, which I’d later realize was very on-brand for how this whole team operates.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ivQ76idpBaPjmnxTh1R6Gg.png" /><figcaption>OA Invite — Litmus</figcaption></figure><h3>The OA — where things got genuinely interesting</h3><p>This is the part of the process I keep coming back to when I tell people about it. Litmus is an <strong>AI-native technical hiring platform</strong> — which means the way they <em>evaluate</em> candidates is itself a reflection of the product philosophy they’re building. It’s not a company bolting AI onto a traditional hiring flow; the assessment format itself was clearly designed around the assumption that working with AI tools is now a core engineering skill, not an accessory to it.</p><p>Instead of a typical coding test with a problem statement and a submit button, I was given a command to fork a repo directly and had to commit changes into it. Before I could even start, I had to give Litmus’s tooling access to my own system so it could install whatever it needed to run and track the assessment locally.</p><p>And then everything was tracked. Not just my final code — <strong>every command I ran, every prompt I sent to an AI model, and which model I chose to use</strong>, all of it fed into their evaluation pipeline. Every commit message, every prompt, every small decision along the way mattered, because none of it was invisible to them.</p><p>That reframed the whole exercise for me. In a traditional OA, the only thing that matters is whether your final solution passes the test cases. Here, the process <em>was</em> the signal. The underlying question wasn’t “can you solve this problem” — it was “<strong>can you actually work effectively with AI as a collaborator</strong>,” which, if you think about where software engineering is heading, is probably a more honest question to be asking candidates in the first place.</p><p>Once I was done, I submitted the whole thing directly from my terminal — no separate submission form, no upload step. It closed the loop on how self-contained and tool-native the entire experience was.</p><h3>The interview — with Litmus’s CTO</h3><p>After the OA, I got an invite to interview with <strong>Litmus’s CTO</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eoWZt7bq4eU_xIqpXdNaGQ.png" /><figcaption>CTO Round — Litmus</figcaption></figure><p>She opened the conversation by introducing herself properly — what they’re building, when they started, and the story behind it — rather than jumping straight into grilling me. It set a tone early on that this wasn’t going to be a rapid-fire technical interrogation.</p><p>Then I introduced myself, and we got into my background and open-source journey — the kind of context that usually only comes up if an interviewer is genuinely curious rather than just checking a box.</p><p>From there, the conversation moved into the OA itself: what I’d actually built, how I’d approached the problem, what was going through my head at each step, and — this part I found particularly useful — what other approaches might have worked better. It wasn’t just “did you get it right,” it was a genuine back-and-forth about tradeoffs and alternate paths I hadn’t considered.</p><h3>Beyond the formal interview</h3><p>Once the structured part wrapped up, the conversation didn’t just end — it opened up. A good chunk of it was around <strong>YC itself</strong>: what the experience has actually been like for them, what changes internally once you’re admitted into the program, and how fast-paced building at a YC-backed startup really is day to day. Not the polished, PR version of “startup life” — an honest one.</p><p>We also ended up talking about AI more broadly. This happened to be right around the time <strong>Claude Code’s source code leaked</strong> online, so naturally that came up too, and we spent some time just discussing what it meant and how it was being talked about.</p><p>What stood out most to me, more than any single topic, was how the conversation <em>felt</em>. She made me genuinely comfortable — comfortable enough that I was sharing my actual thinking rather than reciting something I’d memorized from a textbook or read somewhere online the night before. That’s a surprisingly rare quality in an interviewer, and it changes the entire texture of the conversation. It stopped feeling like an evaluation and started feeling like two people talking about problems they both found interesting.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/526/1*Z9V71UNSLADTLnyCH-LHnw.jpeg" /><figcaption>Y Combinator</figcaption></figure><h3>The outcome</h3><p>As of about <strong>8 months</strong> after this interview, Litmus still hasn’t hired anyone for the role. My best guess — and it is just a guess — is that they’re deliberately moving fast with a small, tight team, heads-down and focused on getting to Demo Day rather than expanding headcount right now. That’s a fairly normal rhythm for an early-stage YC company, even if it means the hiring loop itself stays open longer than usual.</p><p><strong>Verdict</strong>: Don’t Know</p><p><strong>LinkedIn:</strong> <a href="https://linkedin.com/in/divyanshu-rajput8171">https://linkedin.com/in/divyanshu-rajput8171</a><br><strong>Portfolio:</strong> <a href="https://divyanshu-rajput.vercel.app/">https://divyanshu-rajput.vercel.app</a></p><h3>What I took away</h3><p>I’m genuinely grateful for the opportunity and for how the Litmus team handled the whole process, end to end. No offer came out of it, but that’s honestly not the part I remember most. What I remember is an OA that actually tested something real, and a conversation with a CTO who made space for honest thinking instead of rehearsed answers. As a first look at what interviewing with a YC company actually feels like, I couldn’t have asked for a better one.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bbffc63fc2ab" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Interview Experience: Osvi.ai (FDE Intern, Onsite Bangalore)]]></title>
            <link>https://medium.com/@divyanshurajput709/my-interview-experience-osvi-ai-fde-intern-onsite-bangalore-49c3abd9e8b0?source=rss-0942ae804bbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/49c3abd9e8b0</guid>
            <category><![CDATA[voice-ai]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[openai]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[sarvam-ai]]></category>
            <dc:creator><![CDATA[Divyanshu Rajput]]></dc:creator>
            <pubDate>Tue, 15 Sep 2026 13:51:45 GMT</pubDate>
            <atom:updated>2026-09-15T13:51:45.908Z</atom:updated>
            <content:encoded><![CDATA[<p>This was my <strong>first-ever interview in the voice AI space</strong> — even before Zomato/Nugget.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*a1YAqesIscM1eND4YyrD_Q.jpeg" /><figcaption>OSVI AI</figcaption></figure><h3>How it started</h3><p>I’ve been genuinely interested in voice AI for a while, and started digging into the startups building on top of the bigger voice AI providers — companies working with <strong>Sarvam, Smallest.ai, Rumi.ai, and ElevenLabs</strong> to build solutions for other businesses. That’s how I found Osvi.ai.</p><p>I reached out to the founder directly on LinkedIn. He was kind enough to have a conversation with me, but at the time they were still early-stage and not hiring.</p><p><strong>3–4 months later</strong>, he messaged me back out of nowhere, saying they now had openings and I could apply through their portal — and that if my profile looked good, they’d reach out for next steps. So I applied.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*56-SRfNYUAV5Mlhndo0nxw.png" /><figcaption>OA Mail — Osvi AI</figcaption></figure><h3>The OA</h3><p>The next step was an OA — my first time taking this specific kind of assessment. It wasn’t a typical DSA test; it was built around making a <strong>no-code/low-code AI system more reliable</strong>, so prompting skills mattered as much as anything else. It had multiple sections looking at the problem from different angles — the customer’s point of view and the organization’s point of view.</p><p>What stood out: the OA platform itself was <strong>Osvi’s own product</strong>. It was genuinely well-built, and they even sent free tokens over email as a thank-you for completing the assessment.</p><h3>Round 2— with the co-founder</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MF8Hzy5D60PrraVCcgAG3A.png" /><figcaption>Round 2-FDE Intern Role</figcaption></figure><p>I got invited to R1 with the same co-founder I’d originally reached out to on LinkedIn. It started with a warm intro on both sides, and honestly felt more like a <strong>conversation than a coding interview</strong> — about 10–15 minutes covering what they’re building and how they think about solving problems for their customers.</p><p>From there we moved into my background — my open-source work and my competitive programming journey. By the end, he felt I leaned more toward an <strong>SDE profile than FDE</strong>, and set up another round with the co-founder/CTO to dig deeper.</p><h3>Round 2 — with the CTO</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fZmOI-5uynGR-bV4YHaEiQ.png" /><figcaption>Round 1 — SDE Intern</figcaption></figure><p>This round opened with a warm intro too, then went straight into a deep dive on <strong>QuickDraw</strong>, one of my projects (you can check it out on my portfolio linked below).</p><p>It was purely project discussion, and shorter than expected — about <strong>20 minutes</strong> — because the CTO had urgent work come up mid-interview. Instead of cutting the evaluation short, he gave me a task: make some specific changes to my own project. I submitted that assignment a day later.</p><h3>The outcome</h3><p>I didn’t hear back for about a week. I followed up, and learned they’d gone with someone whose profile was more aligned with their immediate business needs.</p><p><strong>Verdict: Rejected.</strong></p><h3>What I took away</h3><p>I’m genuinely thankful to the founder for even giving me the chance — this process opened up the voice AI space for me in a real way. Even after the rejection, I kept learning from it: I ended up implementing a full <strong>STT/TTS pipeline myself using LiveKit and Deepgram</strong>, just to build a deeper, hands-on understanding of how these systems actually work under the hood.</p><p><strong>LinkedIn:</strong> <a href="https://linkedin.com/in/divyanshu-rajput8171">https://linkedin.com/in/divyanshu-rajput8171</a><br><strong>Portfolio:</strong> <a href="https://divyanshu-rajput.vercel.app/">https://divyanshu-rajput.vercel.app</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=49c3abd9e8b0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Interview Experience: Keploy (Two Attempts)]]></title>
            <link>https://medium.com/@divyanshurajput709/my-interview-experience-keploy-two-attempts-be714633aedb?source=rss-0942ae804bbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/be714633aedb</guid>
            <category><![CDATA[dsa-problem]]></category>
            <category><![CDATA[gsoc]]></category>
            <category><![CDATA[keploy]]></category>
            <category><![CDATA[lfx-mentorship]]></category>
            <category><![CDATA[interview]]></category>
            <dc:creator><![CDATA[Divyanshu Rajput]]></dc:creator>
            <pubDate>Mon, 14 Sep 2026 06:11:01 GMT</pubDate>
            <atom:updated>2026-09-14T06:11:01.736Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/654/1*cMCRe7jjygzPPI7sQS2yfQ.png" /><figcaption>Keploy</figcaption></figure><p>Keploy holds a special place for me — it was my <strong>first-ever interview experience</strong>, and I ended up going through the process with them twice, about a year apart.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3s8yt6mvD5Sc7I3JjklyOA.png" /><figcaption>First Attempt Interview — Keploy</figcaption></figure><h3>How it started</h3><p>I was part of the Keploy fellowship, and applied directly through their portal. That’s what got me the shot at an interview. I got a call from HR saying I’d been shortlisted — and the interview was scheduled for the very next day. From what I gathered, that’s just how Keploy tends to operate: they move fast, and you often don’t get much lead time to prepare. Lesson learned — with a process like this, you have to stay interview-ready all the time, not just when you know one’s coming.</p><h3>First attempt (R1)</h3><p>The round was taken by an SDE1 who’d actually done GSoC under Keploy herself, then joined as an intern, and eventually converted to full-time.</p><p>It opened with an intro, then a quick resume discussion — about 5 minutes — before she moved straight to a DSA problem, with the plan to come back to the resume if time allowed later.</p><p>The problem was a <strong>medium-hard, DP-based problem</strong>, something related to subsequences (the exact details are fuzzy now, since this was about a year and a half before I’m writing this). We worked through it together for about half an hour, but I wasn’t able to fully solve it — my DSA fundamentals weren’t where they needed to be at that point.</p><p>After the technical part, she was generous enough to share her own experience working at Keploy and her GSoC journey with them, which was genuinely interesting to hear.</p><p>I didn’t clear this round.</p><p><strong>Verdict </strong>— Rejected</p><h3>Second attempt, a year later</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oAW_utP--9Ng7vN9ERn98Q.png" /><figcaption>Second Attempt Interview — Keploy</figcaption></figure><p>I got another shot at Keploy about a year after the first attempt. Same warm intro to start.</p><p>This interviewer was also GSoC and LFX qualified. We didn’t spend much time on background this time — he moved fairly quickly into a problem, an <strong>advanced graph problem</strong> pulled from the NeetCode 150 sheet.</p><p>I actually cracked the approach and started building out the solution, but ran out of time before I could finish coding it end-to-end. Close, but not quite there within the time limit.</p><p>Rejected again after R1.</p><p><strong>Verdict </strong>— Rejected</p><h3>What stuck with me</h3><p>Even with two rejections, both experiences were genuinely good ones. The Keploy team is warm and easy to talk to — it never felt like a high-pressure interrogation, more like a conversation with someone who actually wanted you to do well. If anything, it’s part of why I keep coming back.</p><p><strong>LinkedIn:</strong> <a href="https://linkedin.com/in/divyanshu-rajput8171">https://linkedin.com/in/divyanshu-rajput8171</a><br><strong>Portfolio:</strong> <a href="https://divyanshu-rajput.vercel.app/">https://divyanshu-rajput.vercel.app</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=be714633aedb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Interview Experience: Zomato Hyperpure Team (SDE Intern) — Eternal]]></title>
            <link>https://medium.com/@divyanshurajput709/my-interview-experience-zomato-hyperpure-team-sde-intern-eternal-ffe6cadcb820?source=rss-0942ae804bbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/ffe6cadcb820</guid>
            <category><![CDATA[interview]]></category>
            <category><![CDATA[blinkit]]></category>
            <category><![CDATA[food]]></category>
            <category><![CDATA[eternal]]></category>
            <category><![CDATA[zomato]]></category>
            <dc:creator><![CDATA[Divyanshu Rajput]]></dc:creator>
            <pubDate>Mon, 14 Sep 2026 05:20:29 GMT</pubDate>
            <atom:updated>2026-09-17T19:53:54.585Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8H2cd_wHb2QHx_MzeFegqw.png" /><figcaption>Invite From Zomato</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/447/1*bx4lQ0SaMt3yvxyZyp1XYg.png" /><figcaption>Hyperpure by Zomato</figcaption></figure><h3>How it started</h3><p>I got a call from HR letting me know I’d been shortlisted. Before anything else, they checked my availability — how many months out I could join — and scheduled the interview around that.</p><h3>Round 1</h3><p>R1 opened with a warm, quick intro. From there, we actually spent a good amount of time on my open-source journey and the work I’ve done there, before ever getting to the resume itself.</p><h3>The DSA round</h3><p>This was the toughest part of the interview — a competitive-programming-style problem based on an actual real-world problem Hyperpure deals with, centered on <strong>graph optimization techniques</strong>. It was genuinely hard. I started working through it but wasn’t able to fully break the problem down.</p><h3>System design</h3><p>After that, we moved to system design — specifically one of my own projects with a fairly large codebase, fully open-sourced and functional, where I’d followed a number of standard techniques and implementation patterns. We spent close to an hour going deep on the design decisions there.</p><h3>The outcome</h3><p>Because I wasn’t able to solve the DSA problem, the interviewer wasn’t able to move my profile forward this time. That said, he was encouraging about my HLD understanding and grasp of design patterns, and the clear suggestion was to put more focused effort into DSA going forward.</p><h3>Beyond the technical rounds</h3><p>We also talked a bit about how the Hyperpure team works day-to-day. The interviewer shared his own path — he’d joined as an intern, converted to SDE1 after about 8–9 months, and has now been at Zomato for roughly 1.5 years total. Hearing that perspective was a good way to close out the round, even with the outcome not going my way.</p><p>The interview ended after R1.</p><p><strong>Verdict</strong>: Rejected</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/738/1*llS9Y-gJrbifakjNNPKLxw.jpeg" /><figcaption>Zomato — Food Delivery Platform</figcaption></figure><p><strong>LinkedIn:</strong> <a href="https://linkedin.com/in/divyanshu-rajput8171">https://linkedin.com/in/divyanshu-rajput8171</a><br><strong>Portfolio:</strong> <a href="https://divyanshu-rajput.vercel.app/">https://divyanshu-rajput.vercel.app</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ffe6cadcb820" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[My Interview Experience: FDE Intern Role at Nugget (Zomato’s Voice AI Team) -Eternal]]></title>
            <link>https://medium.com/@divyanshurajput709/my-interview-experience-fde-intern-role-at-nugget-zomatos-voice-ai-team-eternal-5293dd910139?source=rss-0942ae804bbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/5293dd910139</guid>
            <category><![CDATA[eternal]]></category>
            <category><![CDATA[zomato]]></category>
            <category><![CDATA[interview]]></category>
            <category><![CDATA[fde]]></category>
            <category><![CDATA[voice-assistant]]></category>
            <dc:creator><![CDATA[Divyanshu Rajput]]></dc:creator>
            <pubDate>Sat, 12 Sep 2026 12:43:04 GMT</pubDate>
            <atom:updated>2026-09-12T14:48:27.069Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/148/1*S3FPY0Qab3HHu1hBa7SCoA.png" /><figcaption>Eternal</figcaption></figure><h3>How it started</h3><p>This one started with a vouch. Someone from inside Zomato referred me for the FDE (Forward Deployed Engineer) role, and that’s actually how I learned something about Zomato’s hiring: they don’t run open “careers” postings for this kind of role. If someone inside the org vouches for you, that’s most likely how you get the interview call in the first place.</p><p>The role sits in Zomato’s Voice AI team, under <strong>Nugget</strong> — the voice AI vertical Zomato has built internally. So my interview process was really an interview for Nugget.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/200/1*clUZ6Ouaqv3EAPw25MtqIA.jpeg" /><figcaption><strong>Nugget by Zomato</strong></figcaption></figure><h3>Round 1 — the background round</h3><p>R1 was taken by an SDE2 and ran about <strong>1 hour 45 minutes</strong>. It started with a warm intro, then went deep into my background — and a good chunk of the round was the interviewer trying to understand <em>why</em> I wanted an FDE role at all, given that my profile leaned more naturally toward SDE or AI roles. That question came up more than once, and I had to actually make the case for why FDE was the right fit for me, not just a fallback.</p><p>One thing I really appreciated: Zomato has this practice where the interviewer tells you the result themselves, right at the end of the round, along with specific feedback on what to improve. No waiting days to hear back on how it went. I got positive feedback and moved forward.</p><p>About 10 days later, I got a call confirming I’d cleared R1, and R2 was scheduled 3 days out.</p><h3>Round 2 — the systems round</h3><p>R2 was taken by the <strong>Engineering Manager</strong> and ran about <strong>2 hours 30 minutes</strong> — noticeably longer and heavier than R1, but the interviewer was just as friendly.</p><p>We started with intro-style questions again, then moved fully technical: agentic workflows, HLD-style system design, and a fair amount of discussion around voice AI implementations — RAG pipelines, how LLMs work under the hood, and how I’d think about helping Nugget’s clients integrate these systems. Nugget’s dominant client is Tata 1mg, and a lot of Nugget’s actual work is building and supporting solutions specifically for how their clients interact with the platform, so a lot of the conversation circled back to that.</p><p>From there it pivoted into a deep dive on my own projects — the technical decisions behind them, what I’d used and why.</p><p>This round also included a case-study/puzzle segment. I was given a puzzle with 2 minutes to think it through — I ended up spending closer to 5 minutes on it, didn’t fully solve it, but had a good ~10-minute back-and-forth with the interviewer working through it together.</p><p>We also ended up talking about some of Zomato’s internal products — I’d come across a few of them through Zomato’s public GitHub, including some mistakes their team had made along the way. The interviewer was genuinely open about discussing those, and we spent a good amount of time on it.</p><p>I walked away with what I’d call “middle” feedback — the interviewer seemed satisfied, but I genuinely couldn’t tell if that meant I’d get a callback or not.</p><h3>Round 3 — the onsite in Gurugram</h3><p>This one moved to Zomato’s office in Gurugram. I got there 30 minutes early, and HR was incredibly welcoming — took me to the cafeteria before the interview, where I grabbed breakfast.</p><p>R3 was originally supposed to be with the same Engineering Manager from R2, but he had meetings, so an SDE2 stepped in instead.</p><p>The round opened with resume-based questions — JS, React, AWS — then a voice AI deep dive, and moved into workflows I could imagine building for Zomato using Nugget. The core of the round was a scenario: <em>think of yourself as an organization like Zomato, automating your own customer support team with AI — to the point where you can remove about 80% of the human support team.</em> We spent roughly <strong>1.5 hours straight</strong> on this one scenario, going deep on STT, VAD, TTS models, routing logic, and how all of it ties together.</p><p>At the end of the round, I was told directly that I hadn’t cleared it. The feedback was specific and useful: spend more time on agentic workflows, and reapply in 2–3 months.</p><p><strong>Verdict </strong>— Rejected</p><h3>Overall</h3><p>Three rounds, three different interviewers, and every single one gave me direct, specific feedback in real time — which is rarer than it should be. I didn’t get the offer this time, but I came out of it with a much clearer picture of what an FDE role at a voice AI product like Nugget actually expects, and exactly where I need to level up before I try again.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/738/1*SJLm07Io0MOW2tZs35diyQ.jpeg" /><figcaption>Zomato—Food Ordering Platform</figcaption></figure><p><strong>LinkedIn:</strong> <a href="https://linkedin.com/in/divyanshu-rajput8171">https://linkedin.com/in/divyanshu-rajput8171</a><br><strong>Portfolio:</strong> <a href="https://divyanshu-rajput.vercel.app/">https://divyanshu-rajput.vercel.app</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5293dd910139" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[One Micro-Commit at a Time: My GNOME Journey and OOSC 4.0]]></title>
            <link>https://medium.com/@divyanshurajput709/one-micro-commit-at-a-time-my-gnome-journey-and-oosc-4-0-9b36236c4066?source=rss-0942ae804bbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/9b36236c4066</guid>
            <category><![CDATA[migration]]></category>
            <category><![CDATA[ubuntu]]></category>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[gnome]]></category>
            <category><![CDATA[open-source]]></category>
            <dc:creator><![CDATA[Divyanshu Rajput]]></dc:creator>
            <pubDate>Sat, 12 Sep 2026 06:53:24 GMT</pubDate>
            <atom:updated>2026-09-18T13:36:46.765Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hYLV-F_cc0Yp1T-dDo5Myw.jpeg" /><figcaption>OOSC 4.0 (IIIT Allahabad, India)</figcaption></figure><h3>How I found GNOME</h3><p>I got into GNOME almost by accident. In October 2025, I connected with Aryan Kaushik, a member of GNOME’s internship committee who happens to be from the same city where I’m doing my engineering. That one conversation gave me a direction, where to look, what to explore, and how newcomers actually get started in a project as big as GNOME.</p><p>Then I found one of my alumni (Shobha Tyagi Mam) as a Dec 2013-Mar 2014 Outreachy (formerly Outreach Program Women) GNOME Intern and GNOME Foundation member 2012–2022. She guided me all through the way.</p><p>Following that thread, I started poking around different repos and eventually landed on <strong>gitg (</strong>GNOME’s Git client), which had an open migration project: moving the entire codebase from GTK3 to GTK4.</p><h3>The migration</h3><p>I didn’t do this alone. Two other fellows, Shivam and Sajal, together we built the migration on top of an existing MR from Alberto Fanjul, a GNOME community member and maintainer of gitg who’d already laid some groundwork.</p><p>Early on, we shaped the whole project: migrate in <strong>micro-commits</strong>. Instead of one giant sweeping change, we broke the work into small, reviewable pieces and squashed a few of them later once things stabilized, just enough to keep the history clean without losing the granularity that made review possible in the first place.</p><p>It wasn’t smooth the whole way through. The hardest part, honestly, was keeping the branch properly rebased. Staying consistent across a migration that stretched over months, while GTK4 changes and other MRs kept landing upstream, was a constant balancing act. I ended up with 20+ merged MRs by the end.</p><p>If I had to boil down the one lesson from this whole project, it’s this: <strong>don’t migrate large codebases in a single PR.</strong> Maintainers don’t have the bandwidth to review a wall of changes — and expecting them to is how good migrations die in review queues. Small, easy-to-review, easy-to-approve pieces are what actually get merged.</p><h3>Why I wanted to speak</h3><p>This year I attended GUADEC 2026 remotely, and it changed how I saw the whole ecosystem — the scale of it, the people behind it, and how much of GNOME’s progress comes from individuals just showing up and doing the work. It planted the idea that I wanted a stage of my own to talk about GNOME.</p><p>The final push came from watching Aditya from the GNOME Nepal team — his work made me want to go do the same thing: get up, share what I’d learned, and hopefully get someone else excited enough to start contributing too.</p><p>That’s how I ended up applying to speak at <strong>OOSC 4.0</strong>, an international conference organized by OpenPrinting and led by Till Kampetter, held at IIIT Allahabad. I’d actually first heard about OOSC from a friend who’d previously been a GSoC mentee at OpenPrinting — small world.</p><h3>The talk</h3><p>My talk was titled <strong>“One Micro-Commit at a Time: Migrating Legacy GNOME Apps to GTK4.”</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uDI0j3UwLrYtSnr2xUc4Xg.jpeg" /><figcaption><strong>One Micro-Commit at a Time — </strong>Migrating Legacy GNOME Apps to GTK4</figcaption></figure><p>I walked through the whole gitg migration — how we approached a large legacy codebase, what worked, what didn’t, and the rebase struggles that ate up more time than I’d like to admit. I was upfront about the mistakes too: migrating in one big PR is <em>not</em> the way to do it, and I said so plainly, because I think more people need to hear it before they make the same mistake.</p><p>I also talked about the human side of it — how getting involved in something like a GSoC-adjacent effort completely changed how I learn, and how much I’ve gotten just from being around people in the GNOME community who know far more than I do.</p><p>And I made a point of saying this clearly: <strong>not getting into GSoC doesn’t put you behind.</strong> I didn’t get in either. Open source was never really about the stipend or the program name on your resume — it’s about learning from the people around you and giving something back. That’s the part I most wanted people in the room to walk away with.</p><h3>The Feedback</h3><p>The response was better than I expected. A good number of students and other open-source contributors came up afterwards saying the talk had genuinely pushed them to want to start contributing — which, honestly, is the entire reason I wanted to speak in the first place.</p><p>I also got to sit down with Till Kampetter for about half an hour, talking through GNOME and OpenPrinting. That conversation is where I learned about an unfinished project in GNOME Settings for CUPS that still needs someone to pick it up — something I’m seriously considering.</p><p>Beyond that, most of my time at OOSC 4.0 went into conversations with fellow speakers and students there, comparing notes on our respective corners of GNOME, and there was a recurring theme: every project, everywhere, needs more maintainers and more contributors. It’s not a gitg problem or an OpenPrinting problem — it’s an open-source-wide problem, and it made me appreciate how much any single person showing up actually matters.</p><h3>What’s next</h3><p>I recently applied to UbuCon India, and if that comes through, I’m hoping to get a chance to talk about GNOME on that stage too. One talk led to the next connection, which led to the next idea — that seems to be how most of this works.</p><p>If you’re on the fence about starting somewhere in open source: you don’t need a GSoC acceptance letter to matter. Find one repo, one maintainer willing to point you in a direction, and start small. That’s genuinely all it took for me.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_K-hBWrZwXySOCdJAs7J_Q.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9b36236c4066" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[“The Hidden Cost of AI Trends: How Ghibli-Style AI Art is Hurting the Environment”]]></title>
            <link>https://medium.com/@divyanshurajput709/the-hidden-cost-of-ai-trends-how-ghibli-style-ai-art-is-hurting-the-environment-7d6cea0236c6?source=rss-0942ae804bbe------2</link>
            <guid isPermaLink="false">https://medium.com/p/7d6cea0236c6</guid>
            <dc:creator><![CDATA[Divyanshu Rajput]]></dc:creator>
            <pubDate>Tue, 01 Apr 2025 08:44:18 GMT</pubDate>
            <atom:updated>2025-04-01T08:44:18.683Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3C2RqmMq87cqgGX_-E9LlA.png" /></figure><h3>Introduction</h3><p>The internet has recently been obsessed with <strong>Ghibli-style AI-generated art</strong>, with millions of users generating stunning visuals inspired by the legendary animation studio. But behind this viral trend lies a <strong>serious environmental problem</strong> — one that we can’t afford to ignore. As OpenAI’s CEO Sam Altman recently admitted, AI image generation has pushed their GPUs to the limit, leading to <strong>excessive energy consumption and resource depletion</strong>. So, is this trend worth the cost?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*igLRe-mXDY3kkDVd81eUCw.png" /></figure><h3>The Massive Compute Demand of AI Image Generation</h3><p>Unlike traditional digital art, which relies on human creativity and minimal power, AI-generated images require <strong>powerful GPUs running complex neural networks</strong>. These models, such as those used by OpenAI’s image generator, consume:</p><ul><li><strong>High amounts of electricity</strong> to run massive computations.</li><li><strong>Cooling systems</strong> that add to energy consumption.</li><li><strong>Server farms</strong> filled with hardware that needs constant upgrades.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DxJ-Jqbhi7Kd0EONAeblrw.jpeg" /></figure><p>Every time someone generates an image, they are tapping into a system that requires <strong>thousands of GPUs</strong>, drawing power from energy grids that often rely on fossil fuels.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hoZq1lGiXNy9m1rRhvu4Bw.png" /></figure><h3>Environmental Impact: A Hidden Disaster</h3><h4>1. Gigantic Carbon Footprint 🌍</h4><p>AI-powered GPUs consume <strong>megawatts of electricity</strong>, leading to <strong>increased carbon emissions</strong>. If this energy comes from <strong>coal or gas-powered plants</strong>, it accelerates <strong>climate change</strong>. A single AI image generation session might seem harmless, but when millions of users participate daily, the damage adds up.</p><h4>2. Rare Earth Metal Consumption 🏭</h4><p>GPUs and other AI hardware rely on <strong>rare earth elements</strong> like lithium, cobalt, and silicon. Mining these resources:</p><ul><li><strong>Destroys ecosystems</strong> through deforestation and soil degradation.</li><li><strong>Consumes vast amounts of water</strong>, further straining the environment.</li><li><strong>Contributes to toxic waste</strong>, leading to long-term pollution.</li></ul><h4>3. Electronic Waste (E-Waste) ♻️</h4><p>AI demands the latest and most powerful hardware. As technology advances, older GPUs become obsolete <strong>faster than ever</strong>, creating tons of <strong>electronic waste</strong> that is often dumped in landfills.</p><h3>OpenAI’s Struggles: “Our GPUs Are Melting”</h3><p>Sam Altman’s tweet revealed that OpenAI had to <strong>limit the number of free image generations</strong> because their GPUs were overwhelmed. This highlights a <strong>fundamental problem</strong> — as AI becomes more accessible, its <strong>environmental costs skyrocket</strong>. With millions of users pushing AI systems to their limits, the industry is forced to <strong>expand data centers</strong>, further increasing its carbon footprint.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TGhnP2ezUAODm1_tWUvKJw.jpeg" /></figure><h3>The Bigger Question: Is AI Art Worth It?</h3><p>While AI-generated Ghibli-style images are beautiful, <strong>at what cost?</strong> The environmental damage caused by energy-intensive AI models is a pressing issue. Should companies impose stricter limits on AI image generation? Should users be more mindful of how often they generate images?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f7lI9xNH0P_WxbdgzpqlHg.png" /></figure><h3>Possible Solutions</h3><ul><li><strong>Transition to Renewable Energy:</strong> AI companies must <strong>power data centers with solar, wind, or hydro energy</strong> instead of fossil fuels.</li><li><strong>More Efficient AI Models:</strong> Researchers should focus on <strong>reducing the computational cost</strong> of AI models.</li><li><strong>User Awareness:</strong> People should be educated about <strong>the hidden environmental impact of AI trends</strong>.</li><li><strong>Stronger Regulation:</strong> Governments and organizations should enforce <strong>sustainable AI policies</strong>.</li></ul><h3><strong>According to Perplexiety ,</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Gj6jVgoYBofTWZHgrCtdVg.png" /></figure><h3>Conclusion</h3><p>AI is revolutionizing the way we create, but viral trends like Ghibli-style AI art expose the <strong>environmental cost of uncontrolled AI usage</strong>. As we continue exploring AI’s potential, we must also address its <strong>real-world consequences</strong> — before it’s too late. The question remains: Is AI art <strong>just a fun trend</strong>, or is it <strong>a silent contributor to environmental destruction</strong>?</p><p><strong>How much i contribute in this destruction ??</strong></p><h3>The Carbon Footprint of Creating This Blog</h3><p>Ironically, while writing this blog to highlight the environmental impact of AI-generated art, I contributed to the very problem I’m discussing. Let’s break it down:</p><p>Generating a single high-quality AI image can take anywhere from a few seconds to a few minutes, depending on the model and server load.</p><p>This process runs on powerful GPUs that consume significant electricity. Estimates suggest that each AI-generated image can consume around <strong>0.05 to 0.5 kWh of electricity</strong>, depending on the complexity.</p><p>If the electricity source is fossil fuel-based, this could translate to <strong>25–250 grams of CO₂ emissions per image</strong>.</p><p>So , think again and use AI.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7d6cea0236c6" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>