<?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 Rohit Bankar on Medium]]></title>
        <description><![CDATA[Stories by Rohit Bankar on Medium]]></description>
        <link>https://medium.com/@rohitbankar?source=rss-caf371709d4d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*KHAxr-baOtb87QYENP-WBQ@2x.jpeg</url>
            <title>Stories by Rohit Bankar on Medium</title>
            <link>https://medium.com/@rohitbankar?source=rss-caf371709d4d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 02:36:45 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@rohitbankar/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[The Flight of a Lifetime]]></title>
            <link>https://rohitbankar.medium.com/the-flight-of-a-lifetime-a6501403ceca?source=rss-caf371709d4d------2</link>
            <guid isPermaLink="false">https://medium.com/p/a6501403ceca</guid>
            <category><![CDATA[dreams]]></category>
            <category><![CDATA[careers]]></category>
            <category><![CDATA[nature]]></category>
            <category><![CDATA[travel]]></category>
            <category><![CDATA[international-student]]></category>
            <dc:creator><![CDATA[Rohit Bankar]]></dc:creator>
            <pubDate>Mon, 16 Nov 2026 05:46:39 GMT</pubDate>
            <atom:updated>2026-11-16T05:46:39.490Z</atom:updated>
            <content:encoded><![CDATA[<p><em>On January 1, 2024, at twenty-three, I left India for the first time. Twenty-odd hours, two unexpectedly lucky window seats, and a chance to finally see the world I had spent my childhood exploring through history books.</em></p><p>As a child, I had a rather unusual fascination with places. Whenever we traveled somewhere in India, I would find myself wondering about the people who had lived there centuries before us. Wasn’t this the place where some famous battle was fought? Was this the palace I’d read about in my seventh-grade history textbook? Had some king, merchant, or traveler once walked along these very roads?</p><p>I loved the idea that seemingly ordinary places could have extraordinary pasts. A fort was never just a fort, a river was never just a river, and a city was rarely just a collection of buildings. There were stories everywhere, and I liked imagining the people who had lived through them. Sometimes, simply recognizing the name of a place from a history book was enough to make an entire trip worthwhile.</p><p>But for all my fascination with the world, I had never traveled outside India. The countries beyond our borders existed mostly in maps, movies, books, and the stories I’d heard growing up. I’d spent years reading about distant civilizations, wars, kingdoms, and trade routes, imagining what those places must have looked like. I had no idea when, or even if, I would ever get to see them.</p><p>And then, on January 1, 2024, I found myself boarding my first intercontinental journey, from Mumbai to Chicago, with a connection in Istanbul.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1qlAzjg-IjvSzX2el6l9Rw.png" /></figure><p>I was finally going to see a little more of the world.</p><p>Although, admittedly, the circumstances were rather interesting.</p><p>I was leaving behind my family, a perfectly stable job, and almost everything familiar to pursue my master’s degree in the United States. I’d taken a substantial education loan, had never visited the country I was moving to, and was placing a rather significant bet on the possibility of a better career, new experiences, and whatever else life might have waiting for me.</p><p>Was it risky? Absolutely. But I was young, full of energy, and perhaps just optimistic enough to believe that things would eventually work out. There was something irresistible about the idea of starting over somewhere completely unfamiliar, even if I couldn’t quite explain what I expected to find there.</p><p>The night before my departure, December 31, 2023, my parents, my brother, and I were busy packing my bags. I still remember the unusual quietness of that evening. There were probably thousands of thoughts running through everyone’s minds, but very few were being spoken aloud. My parents had their doubts, and I could hardly blame them. Yet they supported me, even when the future I was describing to them was mostly a collection of hopes and possibilities.</p><p>The next morning, they dropped me at Chhatrapati Shivaji Maharaj International Airport in Mumbai. I remember our last hug, and the way my parents looked at me before I walked through the departure gate. I could see sadness, nervousness, and excitement together in their eyes, but I don’t think I’ll ever truly understand the gravity of what they were feeling.</p><p>As for me, I felt as though I were about to enter an enormous forest of possibilities. I didn’t know which paths I’d find or where they might lead, but I didn’t necessarily imagine it as a dark forest. There could be beautiful places, unexpected encounters, and experiences I’d never even thought of. Perhaps that was what made the uncertainty so exciting.</p><p>There’s an airport theory I’ve always liked: once someone walks through a departure gate, you never really meet that exact same person again. So much happens between leaving and returning that some part of you inevitably changes. I had no idea then how true that idea might become for me.</p><p>Inside the terminal, I met the people I was going to be living with at Purdue. We had already planned to become roommates, but there is something amusing about meeting people in person when you’re about to share an entirely new life with them in an unfamiliar country. One of them was late, yet somehow arrived with the confidence of someone who was prepared for absolutely anything. Another had been partying with his friends before coming to the airport. I remember finding the contrast rather funny. They seemed like two very different personalities, and looking back, my first impressions of them turned out to be surprisingly accurate.</p><p>It’s strange to think about those moments now. These were people I barely knew at the time, yet they would go on to become my companions, roommates, and some of my closest friends during the years that followed. So many memories that now feel inseparable from my life began with a few introductions at an airport. I suppose we rarely recognize the beginnings of important friendships while they’re happening. At that moment, we were simply a few young people with luggage, flight tickets, and a rather uncertain future.</p><p>Anyway, coming back to the flight. The FLIGHT!</p><p>For someone who had spent years dreaming about seeing the world, I had made one rather questionable decision. I hadn’t booked a window seat. Yes, despite being ridiculously excited about watching entire countries pass beneath me, I had somehow failed to secure the one seat that would allow me to do exactly that.</p><p>But luck was on my side. Somehow, I ended up with a window seat on my first flight, from Mumbai to Istanbul. I’ve always liked believing that when you truly want something, life occasionally finds a way to make it happen. Maybe this was just a coincidence, but I was delighted. The little oval window beside me was about to become my introduction to a world I had spent years imagining.</p><p>We took off from Mumbai at around six in the morning, and I remember watching the city slowly grow smaller beneath us. Mumbai, the city of dreams, was disappearing into the distance, its buildings and roads gradually giving way to the coastline and then the vast Arabian Sea. For a while, I simply sat there looking outside, trying to absorb the fact that I was finally leaving India.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MKDQ9AtXQQzLuM_-y1vR4A.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FOMTBFyGVurJKWFtGrThsg.png" /><figcaption>Mumbai Fading away…. over the blue ocean</figcaption></figure><p>And somewhere over those enormous stretches of water, the excitement began mixing with something else.</p><p>I think <em>scared</em> is the right word.</p><p>For months, leaving India had been an idea, a plan, something I could discuss over dinner or think about before going to sleep. Now it was happening. Everything familiar, my family, my childhood, the places I’d known all my life, was somewhere behind me. I suddenly wondered when I would see it all again, and whether it would feel quite the same when I did.</p><p>I also remember thinking about the idea of <em>home</em>. Until then, home had been such a simple thing. It was where my family lived, where I grew up, where I knew the roads and the people and the little rhythms of everyday life. But what happens when you leave? Does another place eventually become home? Can you belong to two places at once? Or does moving away mean that some part of you will always feel slightly out of place, wherever you happen to be?</p><p>Perhaps these were unusually philosophical thoughts for someone who had barely begun his first international journey. But there is something about watching your country disappear beneath an airplane that makes you think about things you’d never considered before.</p><p>Fortunately, I didn’t have much time to dwell on them, because soon enough, we were flying over Pakistan.</p><p>Now, this was exciting.</p><p>Pakistan was the first country I had ever seen outside India, and I was curious in a way that’s difficult to describe. I’d heard so much about it growing up, from history lessons, newspapers, political discussions, movies, and countless stories. Over the years, I had developed an image of the country entirely through things other people had told me.</p><p>And suddenly, there it was, right beneath me.</p><p>I remember looking at the fields, settlements, and stretches of land, wondering what life must be like for the people living there. Did they farm differently? Were their cities similar to ours? Did their villages look the same? What did an ordinary day look like for someone living in one of those small settlements? Were their lives really that different from ours, or did we have far more in common than the stories we heard growing up might suggest?</p><p>It was fascinating to see a place that had occupied such a large space in my imagination become something so ordinary and real. Somewhere down there, people were probably going about their mornings, working, eating, talking to their families, or worrying about the same little things people everywhere worry about. Meanwhile, I was sitting thousands of feet above them, trying to understand an entire country through an airplane window.</p><p>I don’t think I slept for even a minute. I was glued to the window, watching everything pass beneath me with the excitement of a child who had been shown the world for the first time. Which, now that I think about it, was probably the most accurate description of what was happening.</p><p>Eventually, the landscape began to change. The familiar stretches of land slowly gave way to something completely different, an enormous expanse of desert that seemed to continue forever. I believe we were somewhere over Iran, though I couldn’t have told you exactly where. What I remember is the sand, stretching in every direction, with hardly any visible settlements, roads, or trees.</p><p>As a child, my imagination of deserts had been shaped by all sorts of things. There were stories about ancient kingdoms, castles hidden among the sand, caravans traveling between distant cities, and mysterious oases appearing in the middle of nowhere. And of course, there was <em>Aladdin</em>, which had probably done more to romanticize deserts in my childhood imagination than any geography textbook ever could.</p><p>But seeing an actual desert was a completely different experience. It wasn’t a neatly arranged landscape with an oasis appearing conveniently when the characters needed water. It was enormous, almost unimaginably so, and for what felt like hours, there was very little outside the window except sand.</p><p>I remember wondering how people had ever managed to cross places like these.</p><p>Long before airplanes, modern roads, or even reasonably accurate maps, human beings had traveled between distant civilizations, carrying goods, ideas, languages, and stories across enormous stretches of difficult terrain. I thought about the ancient trade networks connecting India, China, the Middle East, and Europe. The Silk Road had always fascinated me, but looking down at the scale of the landscape made me appreciate those journeys in an entirely different way.</p><p>How did merchants know where to go? How did they find water, navigate through unfamiliar land, or survive weeks of travel under such difficult conditions? I imagined caravans moving slowly across the desert, travelers looking for the next settlement, and people spending months covering distances that we now cross in a few hours.</p><p>There I was, sitting comfortably inside an airplane, occasionally being served food, traveling at hundreds of kilometers an hour over landscapes that must once have seemed nearly impossible to cross. The thought made me admire our ancestors even more. From the earliest tools made of stone to machines capable of carrying hundreds of people across continents, humanity had come an extraordinary distance.</p><p>And then, somewhere in the middle of all this admiration for human civilization, I had a rather unfortunate thought.</p><p>What if our plane crashed here?</p><p>I looked outside again. There was nothing I could see for miles. No towns, no roads, no obvious signs of human life. My imagination, which had been happily occupied with ancient travelers a few minutes earlier, was now wondering how anyone would ever manage to find us if something went wrong.</p><p>Not exactly the sort of thought you want to have while sitting on an airplane.</p><p>But that’s the funny thing about being thousands of feet above the Earth. One moment, you’re admiring the extraordinary achievements of humanity. The next, you’re reminded of just how small and vulnerable we can be in the face of nature.</p><p>Eventually, the desert began to disappear, and I was greeted by another astonishing sight. Snow-covered mountains stretched across the landscape, with valleys and tiny settlements scattered between them. I believe we were somewhere around northwestern Iran or eastern Turkey.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o4csDgosMYfilG7HTkqMNQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SBswku1JGgIwD8OXaNJIRQ.png" /><figcaption>somewhere over turkey</figcaption></figure><p>I was amazed at how dramatically everything had changed. Just a while earlier, I had been looking at an endless sea of sand, and now there were enormous stretches of snow and ice. The mountains looked beautiful, but there was also something intimidating about them. I remember noticing the occasional small settlement and wondering how people managed to live in places like these, surrounded by difficult terrain and what must be incredibly harsh winters.</p><p>Once again, my thoughts returned to the travelers who had crossed these regions centuries ago. Perhaps some of the old trade routes had passed through landscapes like the ones I was seeing. I tried imagining merchants carrying goods through mountains, moving between valleys and settlements, facing weather and terrain that would be challenging even today. It was difficult enough for me to imagine traveling through these places with modern equipment, let alone hundreds of years ago.</p><p>I’d read about these journeys many times, but it was only now that I was beginning to appreciate the world in which they had taken place. A thousand miles is just a number when you read it in a book. Looking down at the mountains and deserts that make up those miles gives it an entirely different meaning.</p><p>And just when I thought the first flight couldn’t get any more interesting, we began approaching Istanbul.</p><p>Or, as the history enthusiast in me preferred to think of it, Constantinople.</p><p>I remember seeing the Bosphorus as we approached the city, and the excitement I felt was difficult to contain. This was the Bosphorus I’d read about so many times, the waterway connecting the Black Sea to the Sea of Marmara and separating the European and Asian sides of Istanbul.</p><p>For years, Constantinople had been a city of stories in my mind. Stories of the Byzantine Empire, the Ottomans, great sieges, battles, fortifications, castles, and the people who had spent centuries fighting for control of one of the most strategically important cities in the world. There were stories of ambition, love, sacrifice, and entire civilizations whose histories had passed through this remarkable place.</p><p>And now, there it was.</p><p>A real city, with real buildings, streets, and people living their everyday lives.</p><p>I remember finding that almost as fascinating as the history itself. Here’s a city that had occupied such an important place in the imagination of historians, rulers, and travelers for centuries, and yet, for someone living there, it was simply home. Somewhere down there, people were probably getting ready for work, making breakfast, or complaining about traffic, completely unaware that a history enthusiast was looking down at their city with the excitement of someone who had just discovered a lost civilization.</p><p>It was a lovely feeling, seeing something I’d known for so long through stories finally become real.</p><p>Landing in Istanbul was another first. For the first time in my life, I had actually arrived in a country outside India. Even walking through the airport felt interesting. The people, the languages, the conversations, and the little differences in how everything worked gave me my first real experience of being somewhere culturally unfamiliar.</p><p>I remember being unusually eager to talk to random people. Looking back, it feels slightly funny, because it was just an airport and everyone was probably trying to get somewhere. But back then, the possibility of meeting someone from a completely different part of the world was exciting enough on its own. I wanted to experience everything, even the ordinary things.</p><p>Soon enough, it was time for the second flight, from Istanbul to Chicago. After everything I’d seen during the first part of the journey, I could hardly imagine what more there might be to discover.</p><p>And, once again, luck was on my side.</p><p>Nobody occupied the window seat in my row, so I was able to sit beside the window for the second flight as well. Two flights, two unexpectedly available window seats, neither of which I’d properly planned for. At that point, I was beginning to think the universe had decided to be particularly generous to me that day.</p><p>As we took off from Istanbul and began flying over Europe, the landscape gradually changed again. This time, what struck me was how familiar some of it looked. There were broad stretches of agricultural land, fields, roads, towns, and settlements scattered across the countryside. After the enormous deserts and snow-covered mountains, the land seemed full of signs of human life.</p><p>By then, we’d been traveling for hours. Our journey had begun around six in the morning in Mumbai, and because we were flying west, we had enjoyed a remarkably long stretch of daylight. For much of the journey, I had been able to watch the changing landscapes below us. But eventually, as we continued across the Atlantic, the light began to disappear.</p><p>I thought perhaps the most interesting part of the journey was behind us.</p><p>I couldn’t have been more wrong.</p><p>At some point during the flight, I looked outside and saw something I don’t think I’ll ever forget.</p><p>I believe we were flying somewhere over Greenland. The sky was dark, the clouds had mostly cleared, and the moon was illuminating an enormous frozen landscape beneath us. I remember the moon looking almost full, its light reflecting off the ice and making the entire scene appear strangely bright against the darkness.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/630/1*EXn5-LJcoYp42T2OTHW7GA.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/335/1*F7KrP1YHcqSUXlA9l7tjVw.jpeg" /><figcaption>(Greenland) .pretty similar to this…</figcaption></figure><p>It was beautiful in a way I’d never experienced before.</p><p>For as far as I could see, there was ice. Mountains, frozen stretches of land, and an almost complete absence of anything familiar. The moonlight made the landscape look as though it belonged to another planet, somewhere far removed from the crowded cities and familiar places I’d grown up around.</p><p>I sat there looking at it for a long time. There was something peaceful about the view, yet also something deeply unsettling. I couldn’t see roads, buildings, or any obvious signs of human life. I wondered how anyone had ever managed to explore regions like these, let alone survive in them.</p><p>And naturally, my mind returned to the same rather inconvenient question it had asked over the desert.</p><p>What if something happened to the plane here?</p><p>Apparently, I had developed a habit of admiring the most beautiful landscapes on Earth and immediately considering the possibility of getting stranded in them.</p><p>But I couldn’t stop looking. The light of the moon falling across the ice, the enormous silence I imagined beneath us, and the sheer scale of the landscape made it one of the most unforgettable sights of my life.</p><p>I had seen photographs of Greenland and other frozen regions before, of course. But photographs have a way of making enormous places look manageable. Looking at them from an airplane, surrounded by darkness, I began to understand how little those images had prepared me for the actual scale of what I was seeing.</p><p>And then I started thinking about the entire journey.</p><p>In a single day, I’d watched Mumbai disappear into the Arabian Sea, seen the landscapes of Pakistan, crossed vast deserts, admired snow-covered mountains, flown over the Bosphorus and the city I’d once known as Constantinople, watched the fertile lands of Europe pass beneath me, and now found myself looking down at an enormous frozen wilderness.</p><p>The same Earth, and yet so many different worlds within it.</p><p>Our ancestors had spent weeks or months crossing landscapes that I had now watched pass beneath me in a matter of hours. Thousands of years of curiosity, struggle, experimentation, and discovery had somehow brought us to a point where a human could sit inside an airplane and cross half the world in a single day.</p><p>There’s something wonderfully strange about that when you think about it.</p><p>Eventually, we flew over what I believed was Canada, where enormous stretches of snow and ice were still visible beneath us. The darkness began to change, and gradually there were more signs of settlements and human activity as we continued toward the United States.</p><p>I was tired by then, certainly. We’d been traveling for what felt like an eternity, and I had spent almost the entire journey staring outside. But I still didn’t want to miss anything. Every time the landscape changed, there was something new to look at, some new thought to entertain, some new place to add to the collection of memories I’d been gathering since leaving Mumbai.</p><p>And then, as we approached Chicago and began descending below the clouds, I saw the city.</p><p>I still remember that view.</p><p>An enormous expanse of lights stretched across the darkness, with roads forming neat patterns and buildings rising above them. The city seemed to continue in every direction, an almost endless arrangement of illuminated streets and carefully laid-out blocks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kc6y4NjndEoICDnbd2lPxA.png" /><figcaption>Gotham City</figcaption></figure><p>Chicago was huge. Much larger than I had imagined, at least from where I was sitting. Thousands of lights, roads, buildings, and neighborhoods, all forming this enormous human landscape beneath us.</p><p>For some reason, the first thing that came to my mind was “<em>Gotham City</em>” XD.</p><p>Perhaps it was the darkness, the enormous buildings, or the way the illuminated streets seemed to stretch endlessly into the distance. Whatever it was, I remember thinking that Chicago looked like something from a movie. After spending the day seeing places I’d only known through history books, it seemed rather fitting to end the journey with a city that reminded me of fiction.</p><p>We landed at O’Hare International Airport at around six in the evening.</p><p>January 1, 2024.</p><p>The same day on which I’d left Mumbai.</p><p>We’d been traveling for roughly twenty-odd hours, crossed continents and oceans, and watched the world change beneath us more times than I could count. Yet somehow, thanks to the time difference of nearly twelve hours, the calendar insisted that it was still New Year’s Day.</p><p>I found that rather amusing. So much had happened since the morning that it felt impossible for everything to belong to a single date. Yet here I was, on the other side of the world, having arrived on the same day I’d left.</p><p>And then the airport doors opened.</p><p>Oh, the cold.</p><p>I had heard about American winters before, but nothing could have prepared me for that first blast of Chicago air. Coming from Mumbai, it felt as though someone had opened the door to an enormous freezer. It must have felt like minus twenty degrees to me, and I remember wondering whether I’d seriously underestimated this particular part of the adventure. And there I was, standing in a country I’d never visited, with no American phone number, no bank account, and an accommodation I hadn’t even seen in person. Apart from my future roommates, I barely knew anyone. There was a university waiting for me, an education loan to repay, and an entire life I would have to figure out, one unfamiliar day at a time.</p><p>I suppose that’s the peculiar mixture of uncertainty and optimism that so many international students carry with them. You leave behind the people and places you know, make sacrifices whose worth you can’t yet measure, and travel halfway across the world believing that something wonderful might be waiting for you. I’d come to America with that same hope, believing that hard work could open doors I’d never imagined. Whether it would all work out, I had absolutely no idea. But standing there, with the cold Chicago air on my face, a few suitcases, some new friends, and a world that suddenly seemed much larger and more interesting than it had the day before, I couldn’t wait to find out.</p><p><em>Sayonara, for now.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a6501403ceca" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Grammarly Interview Experience — Machine Learning Engineer Intern]]></title>
            <link>https://rohitbankar.medium.com/grammarly-interview-experience-machine-learning-engineer-intern-080583d34c8d?source=rss-caf371709d4d------2</link>
            <guid isPermaLink="false">https://medium.com/p/080583d34c8d</guid>
            <category><![CDATA[coding]]></category>
            <category><![CDATA[coding-interviews]]></category>
            <category><![CDATA[internships]]></category>
            <category><![CDATA[data-structures]]></category>
            <category><![CDATA[grammarly]]></category>
            <dc:creator><![CDATA[Rohit Bankar]]></dc:creator>
            <pubDate>Wed, 14 Jan 2026 20:10:30 GMT</pubDate>
            <atom:updated>2026-01-14T20:10:30.874Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Grammarly Interview Experience — Machine Learning Engineer Intern</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Sai31SYgSVX0NFZE.png" /></figure><p>Interviewing at Grammarly felt structured and thoughtful. It wasn’t just “Data Structures”; they tested a mix of <strong>ML understanding, problem-solving under time pressure, and values fit</strong>.</p><p>Every round evaluated a slightly different dimension. The technical rounds rewarded clarity and speed, while the final round placed a strong emphasis on how you work, collaborate, and think about impact.</p><p>This blog is a candid walkthrough of my Grammarly interview journey -from the online assessment to the final behavioral round. I’ll share what went well, what I learned, and how the process was structured, in case you’re preparing for an MLE-style interview loop at Grammarly.</p><h3>Recruiter reaches out</h3><p>After I applied, Grammarly’s recruiting team reached out and explained the process clearly: there would be an <strong>online assessment</strong>, followed by interviews that included a recruiter screen, a technical round, and a final values-focused round.</p><p>The coordination was smooth overall — there was even a quick time-zone clarification since I was in <strong>IST</strong> at that time, hence they adjusted the schedule when an interviewer had a conflict.</p><h3>Timeline</h3><ul><li><strong>Online Assessment:</strong> Nov 18, 2024</li><li><strong>Recruiter Screening Call:</strong> Dec 2, 2024</li><li><strong>Technical Round (DSA):</strong> Dec 11, 2024</li><li><strong>Final Round (EAGER / Behavioral):</strong> Dec 19, 2024</li><li><strong>Decision:</strong> Jan 17, 2025</li></ul><h3>Online Assessment</h3><p>The OA was a mix of ML and CS fundamentals rather than purely LeetCode.</p><p>One part focused on <strong>ML concepts and model-building reasoning</strong> — the kind of questions that check whether you actually understand core ideas (not just terms). Another part leaned into <strong>coding/problem-solving</strong>, including DSA-style questions and a practical ML task where the solution wasn’t completely rigid — your approach and reasoning mattered.</p><h3>Recruiter Screening Call</h3><p>This was a straightforward conversation focused on alignment.</p><p>We talked about my background, what kind of ML work I’m interested in, and what I’m looking for in an internship. It also covered basic logistics and expectations. It didn’t feel like a test — more like Grammarly trying to ensure the role and team match makes sense before moving forward.</p><h3>Technical Interview</h3><p>This round was the most time-pressured part of the process.</p><h3>Structure (approx.)</h3><ul><li>Quick intro</li><li><strong>Three back-to-back DSA questions (Leetcode Medium)</strong></li><li>A few minutes at the end for questions</li></ul><p>The problems were around <strong>medium difficulty</strong>, and the pacing mattered a lot. Even if you know the concepts, you have to stay organized: clarify quickly, communicate your approach, code cleanly, and test without getting stuck in debugging loops.</p><p>What I felt the interviewer was really looking for wasn’t just correctness — it was:</p><ul><li>how quickly you can lock onto a solid approach</li><li>whether your code stays clean under time pressure</li><li>How do you handle edge cases without panicking</li><li>How do you debug and recover if something goes wrong</li></ul><p>This round rewarded “time-bound interview-style practice” more than casual LeetCode practice.</p><h3>Final Round — EAGER Values / Behavioral (Dec 19, 2024)</h3><p>The final round was more conversational and focused on <strong>career &amp; working history</strong> and Grammarly’s <a href="https://www.grammarly.com/blog/company/how-eager-values-shape-our-culture/"><strong>EAGER values</strong></a>.</p><p>It wasn’t the typical “tell me about a time…” rapid-fire format. Instead, it felt like a deeper conversation about how I operate in real situations — ownership, teamwork, communication, and how I handle ambiguity.</p><p>This round mattered a lot. Grammarly clearly cares about values alignment, and you can tell they’re looking for someone who will be a good teammate, not just someone who can solve problems.</p><h3>Decision (Jan 17, 2025)</h3><p><strong>Verdict: Positive feedback, but no hire.</strong></p><p>This was the tough part — I got encouraging feedback from the process overall, but didn’t end up getting an offer.</p><p>Even so, I walked away feeling that the process was fair and that I learned a lot from it.</p><h3>Takeaways</h3><p>This experience reinforced a few things for me:</p><p>Practicing DSA under strict time limits helps, but what really matters is being able to solve problems <strong>quickly while staying structured</strong> — especially when multiple questions are packed into a single round.</p><p>Also, Grammarly’s final round is not a formality. If you treat the values around casually, you’ll miss what they’re really evaluating. Being able to clearly explain how you work — how you communicate, collaborate, and take ownership — genuinely matters.</p><p>Even though I didn’t get an offer, the process pushed me to improve how I handle time pressure, and it reminded me to prepare for behavioral rounds with the same seriousness as technical ones.</p><p>Thanks for reading!</p><p>If this was helpful, feel free to clap, follow, or connect — I’ll be sharing more interviews and engineering experiences soon. Feel free to connect with me on <a href="https://www.linkedin.com/in/rohitbankar/">LinkedIn</a>.</p><p><em>Note: I haven’t shared any exact interview questions because of NDA.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=080583d34c8d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Autodesk Interview Experience — Cloud Infrastructure Engineer Intern]]></title>
            <link>https://rohitbankar.medium.com/autodesk-interview-experience-cloud-infrastructure-engineer-intern-3a62bd12437f?source=rss-caf371709d4d------2</link>
            <guid isPermaLink="false">https://medium.com/p/3a62bd12437f</guid>
            <category><![CDATA[internships]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[cloud-infrastructure]]></category>
            <category><![CDATA[interview-experience]]></category>
            <category><![CDATA[autodesk]]></category>
            <dc:creator><![CDATA[Rohit Bankar]]></dc:creator>
            <pubDate>Sat, 03 Jan 2026 03:06:50 GMT</pubDate>
            <atom:updated>2026-01-03T03:06:50.549Z</atom:updated>
            <content:encoded><![CDATA[<h3>Autodesk Interview Experience — Cloud Infrastructure Engineer Intern</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/417/1*V0mUgU79RPtOz6_y7mfUmw.png" /></figure><p>I went through Autodesk’s interview loop for the <strong>Cloud Infrastructure Engineer Intern</strong> role, and it was one of the more organized and straightforward ones I’ve had. The process was simple and well-structured: one technical round with a coding question and scaling follow-ups, and a final round with the hiring manager focused on my background and values.</p><h3>Timeline</h3><ul><li><strong>Mid-October 2024:</strong> Applied</li><li><strong>November 14, 2024:</strong> Recruiter reached out</li><li><strong>November 18, 2024:</strong> Technical interview (Round 1)</li><li><strong>November 26, 2024:</strong> Hiring manager interview (Round 2)</li><li><strong>December 2, 2024:</strong> Offer decision ✅</li></ul><h3>Round 1 — Technical Interview (Coding + Scaling + Resume)</h3><p>The first round started with a short introduction and then moved into a coding problem.</p><h4>Coding Question (HashMap + Sliding Window)</h4><p>The question was in the <strong>LeetCode easy-to-medium range</strong> and was based on a <strong>hash map + sliding window</strong> pattern. It wasn’t meant to trick you — it was more about whether you can recognize the pattern, implement it cleanly, and handle edge cases.</p><p>I focused on explaining my approach before coding, then implementing step-by-step while calling out things like:</p><ul><li>what the window represents,</li><li>what the hash map is tracking,</li><li>how and when the window expands/shrinks,</li><li>and the time/space complexity.</li></ul><h3>Follow-ups: Scaling and Trade-offs</h3><p>After the solution, the interviewer asked follow-up questions around scaling. The discussion was along the lines of:</p><ul><li>What changes if the input is extremely large?</li><li>How would you approach this if the data is distributed?</li><li>What would become the bottleneck (CPU, memory, network)?</li><li>How would you structure it for large-scale processing?</li></ul><p>This part felt less like “name a tool” and more like showing that you understand the practical side: partitioning data, reducing memory pressure, and thinking about how the algorithm behaves when it’s no longer running on a single machine.</p><h3>Resume Deep Dive (Cloud + ML)</h3><p>The second half of the interview shifted into my resume. The interviewer asked questions around my cloud experience and ML-related work — mostly probing why I made certain design choices, what trade-offs I considered, and how I would approach similar problems in a real production environment.</p><p>Overall, Round 1 was a good mix of coding, systems thinking, and technical communication.</p><h3>Round 2 — Hiring Manager Interview (Resume + Autodesk Values)</h3><p>The second round was with the hiring manager and was more conversational.</p><p>We talked through my background and a few key projects, and then the discussion moved toward <strong>Autodesk values/principles</strong>. The questions were practical and grounded — the kind that aim to understand how you work in a team, how you handle ambiguity, and how you communicate.</p><p>The hiring manager seemed happy with my answers, and the conversation ended on a positive note.</p><h3>Verdict</h3><p><strong>Offer decision on December 2, 2024 — positive outcome ✅</strong></p><h3>Takeaways</h3><p>A few things I felt made the process go smoothly:</p><ul><li>Being comfortable with common DSA patterns like <strong>HashMap + sliding window</strong> helps a lot, especially when you can explain it clearly.</li><li>The scaling follow-ups are important — even if the original question is small, interviewers want to see if you can reason beyond the single-machine solution, and how well you can scale your solution to the level of real-life systems.</li><li>The hiring manager round is not just “behavioral.” It’s about whether your working style fits the team, and whether you can communicate clearly and professionally.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3a62bd12437f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[TikTok Interview Experience — Software Engineer Intern (Privacy & Security)]]></title>
            <link>https://rohitbankar.medium.com/tiktok-interview-experience-software-engineer-intern-privacy-security-3acb0644fd56?source=rss-caf371709d4d------2</link>
            <guid isPermaLink="false">https://medium.com/p/3acb0644fd56</guid>
            <category><![CDATA[sde-interview]]></category>
            <category><![CDATA[data-structures]]></category>
            <category><![CDATA[internships]]></category>
            <category><![CDATA[ti̇ktok]]></category>
            <category><![CDATA[interview]]></category>
            <dc:creator><![CDATA[Rohit Bankar]]></dc:creator>
            <pubDate>Sun, 21 Dec 2025 01:58:29 GMT</pubDate>
            <atom:updated>2025-12-21T01:58:29.713Z</atom:updated>
            <content:encoded><![CDATA[<h3>TikTok Interview Experience — Software Engineer Intern (Privacy &amp; Security)</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*tnipHZ34Mws29Ady.gif" /></figure><p>Interviewing at TikTok is intense. The process moves fast, expectations are high, and even small gaps can quietly decide your outcome. You might feel confident walking out of one round and still realize later that there was very little room for mistakes.</p><p>What stood out to me most was that even when things went <em>well</em>, I walked away knowing exactly what I needed to improve, whether it was speed, depth, or how sharply I understood the systems I talked about.</p><p>This blog is a candid walkthrough of my entire TikTok interview journey from the online assessment to the final technical round. I’ll share what went right, where I slipped, and the practical lessons I took away, especially for anyone preparing for software engineering interviews.</p><h3>Recruiter reaches out</h3><p>I was contacted by a recruiter from TikTok for the <strong>Software Engineer Intern (CIS Privacy &amp; Security)</strong> role for Summer 2025. The email indicated that there will be an online assessment, followed by <strong>two technical interview rounds</strong>.</p><h3>Timeline</h3><ul><li><strong>Online Assessment:</strong> October 5, 2024</li><li><strong>Technical Round 1:</strong> October 28, 2024</li><li><strong>Technical Round 2:</strong> November 5, 2024</li><li><strong>Final Decision:</strong> November 6, 2024</li></ul><h3>Online Assessment</h3><p>The Online Assessment felt demanding but fair.</p><p>Based on previous TikTok interview patterns, online assessments often include MCQs on core CS fundamentals such as databases, networking, operating systems, and data structures, followed by coding problems. In my case, the format was similar; there were five MCQs on CS fundamentals, along with two LeetCode medium-level DSA questions in a standard online coding environment. The questions were manageable. I was able to solve both problems within the time limit and pass all the test cases, which helped me move forward to the interview rounds.</p><h3>Technical Interview 1 (60 minutes)</h3><p><strong>Structure:</strong></p><ul><li>5 min -&gt; Introduction</li><li>25 min -&gt; Coding question (Initial question + Follow-up)</li><li>10 min -&gt; Unit testing</li><li>10–15 min -&gt; Extension on the initial coding question</li><li>5 min -&gt; Room for feedback/questions for the interviewer.</li></ul><p>This round turned out to be one of the better interview experiences I’ve had.</p><p>The interviewer started with a <strong>sliding window–based problem</strong>. At first, it felt quite easy - something along the lines of a standard sliding window question (<a href="https://leetcode.com/problems/minimum-size-subarray-sum/description/">https://leetcode.com/problems/minimum-size-subarray-sum</a>). But as the interview progressed, the interviewer kept adding constraints and variations, gradually increasing the difficulty.</p><p>What stood out here was that the focus wasn’t just on reaching the final answer. The interviewer cared deeply about how I was thinking. I explained my approach step by step, discussed trade-offs, and adjusted my solution as new constraints were introduced.</p><p>Once the solution was done, I was asked to <strong>write unit tests</strong> for it. This was a nice change from the usual “just code and move on” style. It tested how I think about correctness, edge cases, and reliability.</p><p>Toward the end, the conversation shifted to scalability. I was asked how I would handle the same problem if the input data were massive, in the order of terabytes. This turned into a discussion around distributing data across machines and processing it efficiently. I gave a high-level approach using distributed processing ideas and Spark-style computation.</p><p>Overall, this round went well. The interviewer seemed happy with my communication, testing mindset, and ability to reason about scaling.</p><p>Verdict: Passed.</p><h3>Technical Interview 2 (60 minutes)</h3><p><strong>Structure:</strong></p><ul><li>5 min -&gt; Introduction</li><li>40 min -&gt; Coding question</li><li>10 min -&gt; Resume-based question</li><li>5 min -&gt; Room for feedback/questions for the interviewer.</li></ul><p>The second technical round followed a structured format: a short introduction, a long coding session, some resume-based questions, and a few minutes at the end for questions.</p><p>The main coding problem was the classic <strong>LRU Cache (</strong><a href="https://leetcode.com/problems/lru-cache/description/">https://leetcode.com/problems/lru-cache/description/</a><strong>)</strong>.</p><p>I walked through the entire solution in detail, explaining how to combine a hash map with a doubly linked list to achieve constant-time operations. I implemented the solution correctly, but I initially missed one edge case. The interviewer pointed it out, and I fixed it during the interview.</p><p>After the coding portion, the discussion shifted toward systems and big data technologies. I was asked questions related to S3, Delta Lake, Hive, Spark, and Hudi. I was able to discuss most of these comfortably, but I couldn’t answer one question around Hudi in sufficient depth, which I believe played a role in the final decision.</p><p>Verdict: Not moving forward.</p><h3>Takeaways</h3><p>This interview experience taught me a lot.</p><p>Practicing DSA questions under strict time limits made a real difference, but it also showed me that I still need to become faster and more consistent under pressure. Clearly articulating my thought process during interviews helped me a lot, and I also learned how to solve problems more efficiently while explaining my approach.</p><p>For roles that touch privacy, security, and large-scale systems, <strong>deep familiarity with big data technologies is non-negotiable</strong>. Even a single weak area can become a deciding factor.</p><p>While I didn’t move forward, the experience pushed me to sharpen both my DSA speed and my depth in distributed systems. In that sense, it was absolutely worth it.</p><p>Thanks for reading!</p><p>If this was helpful, feel free to clap, follow, or connect — I’ll be sharing more interviews and engineering experiences soon. Feel free to connect with me on <a href="https://www.linkedin.com/in/rohitbankar/">LinkedIn</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3acb0644fd56" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How AI Is Revolutionizing Project Management]]></title>
            <link>https://rohitbankar.medium.com/how-ai-is-revolutionizing-project-management-dab2e2acdebf?source=rss-caf371709d4d------2</link>
            <guid isPermaLink="false">https://medium.com/p/dab2e2acdebf</guid>
            <category><![CDATA[scrum]]></category>
            <category><![CDATA[project-management]]></category>
            <category><![CDATA[management]]></category>
            <category><![CDATA[best-practices]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Rohit Bankar]]></dc:creator>
            <pubDate>Mon, 17 Mar 2025 06:28:34 GMT</pubDate>
            <atom:updated>2025-03-17T06:28:34.426Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*SnGffd5pvohKo-y1" /></figure><p><em>Exploring the AI transformation in IT project management — from task automation to risk prediction and beyond.</em></p><p>In today’s fast-paced business landscape, artificial intelligence (AI) isn’t just a buzzword — it’s a powerful catalyst reshaping industries. Project management, particularly within IT, is experiencing a transformative shift thanks to AI. Drawing from comprehensive research, data, and real-world case studies, this article dives into how AI can automate mundane tasks, optimize resources, and predict risks, ultimately leading to smarter, more efficient project outcomes.</p><h3>The Evolution of Project Management</h3><p>Project management has come a long way — from the manual, labor-intensive methods used to build ancient wonders to today’s sophisticated, tech-driven approaches. Early tools like Gantt charts and Critical Path Methods laid the foundation, but modern software such as Microsoft Project and Agile frameworks have redefined how we plan, execute, and monitor projects. Now, with AI entering the mix, the evolution is reaching a whole new level.</p><h3>AI in Action: Transforming Daily Operations</h3><h4>Task Automation: Liberating Human Creativity</h4><p>One of the most exciting applications of AI in project management is task automation. By automating repetitive tasks, AI allows project managers to focus on higher-level strategic decisions. For example:</p><ul><li><strong>Scheduling and Planning:</strong> AI tools analyze historical data to estimate resource requirements and automatically adjust schedules in real time.</li><li><strong>Progress Monitoring and Reporting:</strong> Platforms like Asana and Smartsheet use AI to compile data from multiple sources, providing managers with real-time dashboards and alerts.</li><li><strong>Enhanced Communication:</strong> AI-powered virtual assistants can schedule meetings, send reminders, and even summarize discussions, ensuring that everyone stays on the same page.</li></ul><h4>Optimizing Resources and Forecasting Budgets</h4><p>Effective resource allocation is key to any project’s success. AI can analyze vast amounts of data to match the right skills to the right tasks, optimize workloads, and even forecast budgets more accurately. This not only improves team efficiency but also prevents project delays and cost overruns.</p><ul><li><strong>Skill Matching:</strong> By assessing employees’ strengths and past performance, AI can suggest optimal team compositions.</li><li><strong>Workload Optimization:</strong> Continuous monitoring helps ensure no team member is overburdened, thus reducing burnout.</li><li><strong>Budget Forecasting:</strong> AI leverages historical project data to provide up-to-date budget predictions, minimizing unexpected financial challenges.</li></ul><h4>Predicting and Mitigating Risks</h4><p>AI’s ability to analyze patterns and predict future outcomes is transforming risk management. Early warning systems powered by AI can detect potential pitfalls well before they escalate into critical issues. For instance, predictive models can flag resource shortages or scheduling conflicts, enabling proactive interventions.</p><ul><li><strong>Early Warning Systems:</strong> AI can trigger timely alerts when it detects anomalies that may indicate future delays or budget issues.</li><li><strong>Risk Prioritization:</strong> By evaluating the likelihood and potential impact of risks, AI helps project managers focus on the most pressing threats.</li><li><strong>Mitigation Strategies:</strong> Drawing on data from similar projects, AI recommends actionable steps to mitigate risks and ensure projects stay on track.</li></ul><h3>Real-World Impact: Case Studies and Success Stories</h3><p>Several industry leaders are already harnessing the power of AI in their project management practices. Platforms like Microsoft Project, Jira, Monday.com, Asana, and ClickUp each bring unique AI-driven features — from timeline predictions and workload balancing to automated risk assessment. These tools have been shown to reduce manual scheduling by up to 60%, improve resource utilization by over 25%, and significantly cut down on timeline estimation errors.</p><h3>Challenges on the Horizon</h3><p>Despite the clear benefits, integrating AI into project management isn’t without its challenges. Legacy systems, data quality issues, and a shortage of skilled personnel can create hurdles in the adoption process. Organizations must invest in change management, employee training, and robust data governance to overcome these obstacles. Moreover, ethical considerations such as data privacy and algorithmic bias must be addressed to build trust in AI-driven solutions.</p><h3>Looking Ahead: The Future of AI in Project Management</h3><p>The future promises even more innovative applications. Trends like explainable AI (XAI) are emerging to provide greater transparency into AI decision-making, while the integration of AI with IoT and blockchain is set to enhance data reliability and security. Furthermore, the development of AI-powered virtual project managers could soon redefine the role of human managers — ushering in a new era of hybrid project management that balances cutting-edge technology with human expertise.</p><h3>Conclusion</h3><p>AI is not just a tool for automation; it’s a game-changer that is redefining project management. With its ability to enhance decision-making, boost operational efficiency, and predict risks accurately, AI is paving the way for more successful and innovative projects. While challenges remain, the potential benefits far outweigh the obstacles. Organizations that embrace AI today are positioning themselves to lead tomorrow’s competitive market.</p><p><em>What are your thoughts on integrating AI into your project management processes? Share your experiences and join the conversation in the comments below.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dab2e2acdebf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Zookeeper- What is it?]]></title>
            <link>https://rohitbankar.medium.com/zookeeper-what-is-it-df7e9bcba9f1?source=rss-caf371709d4d------2</link>
            <guid isPermaLink="false">https://medium.com/p/df7e9bcba9f1</guid>
            <category><![CDATA[zookeeper]]></category>
            <category><![CDATA[consensus-algorithm]]></category>
            <category><![CDATA[distributed-systems]]></category>
            <category><![CDATA[leader-election]]></category>
            <dc:creator><![CDATA[Rohit Bankar]]></dc:creator>
            <pubDate>Fri, 10 Mar 2023 07:00:13 GMT</pubDate>
            <atom:updated>2023-03-10T07:00:13.988Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Zookeeper Overview-</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*AOaYqCq5NLdMGaYt.png" /></figure><p>In this era, no high scale system can survive without distributed architecture. Most of the high-scale applications need to be scaled horizontally as scaling vertically has its limitation. When we scale a system horizontally it has its own set of unique and challenging problems like general consensus, data inconsistency, process synchronization, resource management, fault tolerance, atomicity, etc.</p><p>So in this distributed world, where does Apache Zookeeper fits- As the name suggest it’s a library that allows us to have coordination in distributed systems. Here are some of the distributed system problems that zookeeper solves-</p><ol><li><strong>Leader Election</strong>- In a multi-node cluster, the election can be held by a zookeeper through various algorithms. You can develop your algorithm(discussed in detail later) or use some standard consensus algorithms for leader election like RAFT, ZAB and PAXOS</li><li><strong>Cluster Management</strong>- Work as a registry for clusters, maintains if a server is joining or leaving a cluster, and other miscellaneous info.</li><li><strong>Configuration Management</strong>- It can keep general application configuration that all nodes in a cluster can access.</li><li><strong>Locks in systems</strong> — Zookeeper can help to maintain locks on a shared resource as it helps for mutual exclusivity and prevents data corruption and data inconsistency.</li><li><strong>Synchronization service</strong>- Zookeeper can be used for implementing any custom build synchronization algorithm using its znodes structure(discussed below)</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/0*GR5LtBxs06RZJyRw.jpg" /><figcaption>Client — can be distributed servers(not external clients querying the application)</figcaption></figure><p>Zookeeper can be in itself of distributed nature to have high availability among it, as shown in the image above.</p><h3>Zookeeper architecture-</h3><p>Before diving deep into the complex leader election algorithms it’s important that we understand a simple leader election algorithm and how we can implement it ourselves using zookeeper. To do that we need to have a basic understanding of zookeeper architecture.</p><p>Zookeeper maintains a tree-like data structure for storing values. The nodes in the tree are called znodes. These znodes are similar to folders/files in the Unix file system with some additional abilities(these additional abilities help us to solve many distributed system debacles).</p><p>Some important properties of znodes-</p><ol><li>Stores data and have children znodes(like a typical n-order tree)</li><li>Supports operations like create, delete, getData, exists, setData, etc.</li><li>Supports username/password-based authentication and has a watch feature on znodes(to notify clients about any changes in znodes children or znodes data)</li></ol><p>Different types of znodes-</p><ol><li><strong>Persistent Znode</strong>- These nodes are permanent znodes(as the name suggests), once created they will be there forever unless you manually delete them.</li><li><strong>Ephemeral Znode</strong>- These are temporary znodes, they come and go with the client that created them. After a client goes down or we shut it off all the ephemeral znodes created by the client get deleted by the zookeeper.</li></ol><p><strong>Persistent Sequential Znode</strong>- Same as persistent znode, but only that zookeeper attaches a sequential number as a suffix to it. For ex- <em>pznode1_0001</em>,<em>pz_node2_00002</em>.</p><ol><li><strong>Ephemeral Sequential Znode</strong>- Same as ephemeral znode, but only that zookeeper attaches a sequential suffix to it. For ex- <em>eznode_0001,eznode_0002</em>.</li></ol><p>Sequential Znodes are very useful in leader elections- have discussed in deep below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/442/0*rZ1kGSmnZW7DkotU.jpg" /><figcaption>Zookeeper internal architecture(every node present is a znode)</figcaption></figure><h3>Leader election-</h3><p>Many distributed systems follow the design of leader-follower, in which a server is elected as a leader through a consensus algorithm. Having a leader in a distributed system solves many problems and has huge advantages attributed to it.</p><p>Some of the benefits of having a leader-</p><ol><li>A single place to look for logs and metrics reduces partial failure.</li><li>It can simply tell other servers(followers) about some important changes rather than building consensus.</li><li>It offers clients consistency because they can see and control changes made to the state of the system.</li><li>It maintains all servers in the same state by using consensus algorithms, hence providing atomicity to the client.</li></ol><p><strong>Example of a simple leader election using ephemeral sequential znodes-</strong></p><ol><li>Any server (among all available servers) creates a persistent znode- <strong><em>/election</em></strong><em>.</em> (As only one of them will succeed in creating a persistent znode)</li><li>We will maintain the list of servers as ephemeral znodes(will have data as <em>server.domain.com</em>) under another persistent znode<em> </em><strong><em>/servers.</em></strong></li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZuPmuLyPuTWrqS_JTyPROA.png" /><figcaption>For storing server info</figcaption></figure><p>3. We will use the watch functionality of znodes and add a watch on <strong><em>/election</em></strong><em> </em>znode, such that any addition or deletion to the child nodes of <strong><em>/election</em></strong> will be notified to all the servers with the help of the above znodes.</p><p>4. Now each server that joins the cluster will create an ephemeral sequential znode under the node<strong> <em>/election</em></strong><em>.</em> Data of this ephemeral sequential znode will be their server name (<em>server.domain.com</em>).</p><p>For example, we have 4 servers joining the cluster they will create znodes like this-</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*N57MCjOePYp4hxhw4Va2MA.png" /></figure><p>5. After all the servers complete the creation of znode under <strong><em>/election</em></strong> znode they will call getChildren(“<em>/election</em>”) which will fetch them the data of the least sequenced child node i.e<em> </em><strong><em>/leader-0001</em></strong>. Hence we get the hostname of our leader.</p><p>6. Now in the future, if our leader goes down. Its ephemeral node will also get deleted and as we have set a watch on child nodes of <strong>/election</strong> every server will be notified about it. Then every server will again call getChildren(“<em>/election</em>”) and this time they will get <strong><em>/leader-0002</em></strong> as <strong><em>/leader-0001</em> </strong>was an ephemeral node which will get deleted with the server going down.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kvD9XwHmV4kIvXX-E2cQQw.png" /><figcaption>New leader will be server-2</figcaption></figure><p>This is one of the naive algorithm which we can use in simple distributed systems for leader election.</p><h3>Famous Consensus algorithms for leader election and log replication-</h3><p>I am going to discuss two famous Consensus-based leader election algorithms- RAFT(Reliable, Replicated, Redundant, and Fault Tolerant) and ZAB(Zookeeper Atomic Broadcast).</p><p>Let’s first understand two terminologies- Linerilization and Total Order Broadcast in brief.</p><p><strong>Lineralizability </strong>— It guarantees recency. The basic idea behind it is to make a system appear as if there is always one copy of the data and all the operations are atomic in it. It is the strongest form of consistency</p><p><strong>Total Order Broadcast </strong>— It guarantees reliable delivery and total ordered delivery. Messages are asynchronous, so messages are guaranteed to be delivered but no guarantee about when they will be delivered.</p><p>So if we see things from clients perspectives lineralizability is better than total order broadcasts.</p><h4>RAFT algorithm-</h4><p>Unlike earlier where we consider our node to be a leader or follower, in RAFT a node can be in one of these 3 states-</p><ol><li>Leader</li><li>Candidate</li><li>Follower</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/580/0*U1gCO_FnYeQUDxK4" /></figure><p>Raft seperates its leader election process from log replication.</p><p><strong>Raft’s Leader election algorithm</strong> -</p><ol><li>Every node starts as a follower.</li><li>So the current leader sends regular messages to all the followers in some specific time intervals, this is known as Heartbeat. And every follower has a randomised election timeout defines in it ranging between 150ms — 300ms. So if a follower does not receive Heartbeat from the leader before the election timeout happens, the follower becomes a candidate.</li><li>Every election has a term number, whenever a candidate starts a elections it increases the term number and then starts the election.Now the candidate starts a election with a incremental term number, it votes for itself and requests vote messages from all other nodes(followers).</li><li>Unlike earlier here the request votes contains the term number of election and the candidate’s log. If the candidate which is requesting vote has less updated data in its log than the follower, then also the follower do not vote for the candidate.</li><li>When the candidate receives the majority of the vote(not necessary all of it), It sends a heartbeat message to establish its authority. After receiving the heartbeat all the followers accept the candidate as the leader.</li></ol><p>There may be the case where there are more than one candidates, this can result in multiple leaders, which is a scenario that we should always avoid. So to avoid this, term number comes into the play. If a candidate gets majority and becomes a leader, it sends a heartbeat message which is obviously received by other candidates, other candidates compares there term number with the leaders term number, if there is lower or same they accept defeat and becomes follower and if there term number is greater than the leader they continue there election process(they become a rebel).</p><p>So for a same term there can always be only one leader, if a candidate wins election for a greater term number, then the previous term’s leader become the follower.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/536/0*mcKZFt19REmnxxOp.png" /></figure><p><strong>Raft’s Log replication algorithm-</strong></p><p>RAFT follows the principle of lineralization for log replication.</p><ol><li>Each write or change is redirected towards the leader</li><li>The leader follows 2 Phase commits, in the first phase, he writes the change in its logs and leaves it uncommitted.</li><li>It replicates the value and sends it to all the followers, and it waits for the majority of the followers to write the entry in their logs.</li><li>After receiving the response from the majority of the followers, In the second phase the leader commits the entry and then asks all the followers to commit this value.</li><li>The leader in RAFT ensures that every log of the follower is the same as its log, if some of the followers have fewer logs or more logs, Leader synchronizes its log with the follower forcefully. Then the follower commits the change.</li><li>RAFT implements Lineralization. The cluster now come to a consensus about the system state.</li></ol><h4>ZAB Algorithm-</h4><p>ZAB is very similar to RAFT, hence I will just compare it with RAFT and point out some minor differences.</p><p>Similar to RAFT, ZAB also separates its leader election process from the log replication process.</p><p>Zab leader election algorithm is very similar to RAFT’s, there are just some terminological differences like- the term number is called zxid, their protocol nomenclature is also different, etc. But the core logic is the same.</p><p>ZAB’s log replication also uses 2 Phase Commits like RAFT, but unlike RAFT where lineralization is implemented, in ZAB total order broadcast is implemented. So its logs only require append-only capability.</p><p>There is a special edge case scenario- If the leader fails in the commit phase and at least one of the followers has committed the change, then this follower becomes the leader. Then after being selected as the leader, it lets all other followers commit the change.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/809/0*_CJp4wOYDTrAtWKh" /></figure><p><strong>Uses-<br></strong>RAFT- CoreOS(etcd), Kubernetes(etcd) etc (etcd uses RAFT as its consensus algorithm)<br>ZAB- Apache Kafka, Apache Bookeeper, Apache HDFS, etc (They all use zookeeper where ZAB is implemented)</p><p><strong>PS</strong>-<br>RAFT is considered to be better than ZAB, then also prominent projects like Zookeeper used ZAB because when zookeeper was in the development stage RAFT was not yet released. Hence zookeeper developed its consensus algorithm ZAB. And that’s how ZAB originated.</p><p>I hope this blog helped you to provide a general overview of zookeeper, the internals of zookeeper, leader election and log replication algorithms.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=df7e9bcba9f1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Chess and Life]]></title>
            <link>https://rohitbankar.medium.com/chess-and-life-6fed4138be34?source=rss-caf371709d4d------2</link>
            <guid isPermaLink="false">https://medium.com/p/6fed4138be34</guid>
            <category><![CDATA[life-lessons]]></category>
            <category><![CDATA[chess-lesson]]></category>
            <category><![CDATA[positive-thinking]]></category>
            <category><![CDATA[life]]></category>
            <category><![CDATA[chess]]></category>
            <dc:creator><![CDATA[Rohit Bankar]]></dc:creator>
            <pubDate>Fri, 08 Oct 2021 14:56:52 GMT</pubDate>
            <atom:updated>2021-10-08T14:56:52.521Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/190/1*QPC9mmP0KWd7ZYHIdbPQLQ.jpeg" /></figure><p>From the age of 8 years, I have played chess as a hobby. Chess has taught me many things that are very relevant to live a healthy and successful life. In this blog, I will try to dive deep into the topic of how chess has help me in my life and career and how it can help you!!!!</p><p>Firstly I would like to quote that -”<em> A person who is good at chess will always do very well in terms of career and life</em>”. The reason behind is very simple that there are similar set of skill set that you required to excel at chess as well as life.</p><p>Before diving deep, I would like to introduce myself, and I want to tell everyone that the purpose of this blog is to make everyone aware of how chess has helped me and how it can help everone and why it is a hobby that everyone should have. This blog reflects points from my understanding and experience of how chess has helped me in my life so far and how it can help others. I played chess from my early childhood and represented my school at state and national levels; after joining college for doing BTech(as every other Indian Student :) ), I became head of the chess club at IIT Indore and represented my college on various fronts. This is my short and sweet chess interaction with chess until now, but it has taught me many things that helped me achieve various things in my life.</p><p>Chess is not simply a game, it’s a way of life. Playing a chess game has all components involved in it, which we undergo in everyday life. In chess, we are often under a lot of pressure to make a decision in a limited amount of time with proper evaluation, analysis, strategy, and calculations; that is precisely the mimics of many challenges that we face in our everyday lives. Many scientific studies have shown that children who play chess have improved their IQ level by just playing chess systematically for some months; that’s an astonishing result.</p><p>Some of the greatest life lessons which we can learn from playing chess are-</p><p><strong>Decision Making- </strong>In chess, at every move we make, we are making a decision with calculated risks and benefits after analysing its consequences under time pressure. The same is very much needed in our life, where we should take every decision with uttermost caution, as decisions are powerful and can change the trajectory of your life.</p><p><strong>Flexibility-</strong> In chess, everything does not happen according to our plan; we succumb to surprises from our opponents. Hence, we need to modify or develop an entirely new plan at the very moment for adjusting to the new and unprecedented circumstances. Thus, we have to be very flexible and always be ready to face any situation, which is also the case in our lives, where we never know what is in the store for us. Hence chess helps us to prepare our mindset to face the unexpected.</p><p><strong>Developing logical and analytical Mindset- </strong>In chess, we need to do a lot of calculations for every possible move that our opponent can possibly make; hence a mind of a chess player is highly logical and analytical. A chess player at every moment tries to analyse many moves to find the best possible move. A chess player always takes calculated risks by discarding the worst moves. This exact thing is essential for succeeding in our career and life where we should always take calculated risks and try to think logically and analytically whenever necessary rather than making decisions based on emotions.</p><p><strong>Sacrifices- </strong>Often, in chess, we have to make some sacrifices to win or launch a mating attack. The same is similar in life: to be successful or achieve something in life, you should make sacrifices, ‘nothing is easy in life’ this everyone has heard, but we should understand the gravity of that statement. There is no shortcut in life like in chess. In chess, you cannot afford to make a blunder as one mistake can change everything; the same is very accurate in life.</p><p><strong>Never be comfortable- </strong>Chess players know that you never win till you win. Hence they never get too comfortable even in winning positions; they finish the game and then celebrate. This is also very true in life where you should never expect the result and concentrate on the efforts by giving 100 per cent at the moment.</p><p><strong>Know the system, take action and attack-</strong> This is practical advice which I can bet many of you will not resonate with me, but I found this shockingly accurate. In chess, you first have to understand the game, make an action plan, and finally attack. This is very much true in real life, where we as individuals have minimal scope to stand against the system even if it’s somewhat wrong. Hence, the more logical way to deal is to understand the system around us and utilise it in an ethical but beneficial way. To achieve something in life, we must understand the system to utilise it, make an action plan, and aggressively follow it until we achieve it.</p><p><strong>Great Opening-</strong> A chess player knows the importance of opening as it is the founding stone of a chess game and needs to be strong. The same is in life for every plan we make, every strategy we execute, or every small action we take; we must need to take the first step in the right direction as we know that ‘first impression is the last impression’ and the statement is very true in practical aspects.</p><p><strong>Learn from mistakes-</strong> A chess player never repeats the same mistake twice because of a straightforward reason; he knows that he will lose if he repeats it again. The very same reasoning can be applied to life that if we make the same mistake again, we will get the same undesirable outcome again. But I was astonished that many people simply do seem to overlook the above fact and repeats the same mistake again and again. Hence it is very much essential to learn from our mistakes.</p><p><strong>Bluff-</strong> In chess, there are many situations when we try to bluff the opponent. When we know that our move is not the best move, but the chess player will play it with complete confidence and have facial expressions like everything is right. This skill to bluff and not show your weakness is essential in life because you should never reveal your weakness to the world as there are many people who will try to take advantage of it. It always helps to portray yourself as strong rather than weak in the real world.</p><p><strong>Plan B-</strong> A chess player always has a plan B ready in his mind if his strategy does not work out. In life, you should always have a backup plan, as life can take unexpected turns anytime. Having a backup plan always help you mentally, whether you have to use it or not.</p><p><strong>Big Picture and Creativity- </strong>In chess, we may lose small tactical and maneuvring battles, but our target is always to win the war. The same should be the way we should approach life; we may sometimes fail or not achieve what we desire in an instant, but we should always be focused on our long-term ambitions and goals rather than being upset on small failures. The small failures never define you; instead, you should learn from them and make sure that it never happens again, ultimately leading to achieving your long-term goal.A chess player are always creative as it is the need of the game to be creative if you have to win, similarly in life you have to be creative to get out of some tough situations.</p><p><strong>Thinking ahead- </strong>A good chess player can on an average, always see future 10–12 moves in advance and play his move accordingly. Chess gives us the ability and capability to predict that what can happen in the future. As we know, all the great leaders of our generation are the people who were visionary and accurately predicted the future trends. Hence, if we gain this ability, it will help you succeed in your career and life.</p><p>Playing chess as a profession or playing it as a hobby has numerous benefits included in it, some of which are stated above. There is no exaggeration in saying that chess depicts the miniature version of the life. In this competitive world where everything is a rat race having a hobby of playing chess will definitely help you to give an edge over others; that’s what I have observed in my real life too.</p><p>At last, I would like to comment that whoever you may be or at whatever stage of life you may be, playing chess will surely help in your journey of life in one way or another.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/318/1*x8poaXgss75ELCp6H1yUsQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/341/1*HGOooOXZq3aIVtaQqgIH_g.jpeg" /><figcaption>Extra pic :):) Myself representing IIT indore in Inter IIT Sports Meet.</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6fed4138be34" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Supervised Learning with Scikit-learn]]></title>
            <link>https://medium.com/analytics-vidhya/supervised-learning-with-scikit-learn-df617631216a?source=rss-caf371709d4d------2</link>
            <guid isPermaLink="false">https://medium.com/p/df617631216a</guid>
            <category><![CDATA[regression]]></category>
            <category><![CDATA[classification]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[supervised-learning]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Rohit Bankar]]></dc:creator>
            <pubDate>Sat, 05 Dec 2020 08:58:35 GMT</pubDate>
            <atom:updated>2020-12-09T17:41:19.715Z</atom:updated>
            <content:encoded><![CDATA[<p>So let&#39;s start with what is supervised learning, how is it different from unsupervised learning, what are its practical applications, and how to implement supervised learning with the Scikit-learn library of python.</p><p>Supervised Learning is an approach to create Artificial Intelligence where we provide input and expected output results to the supervised learning model and the model has the task to detect underlying patterns and relationships. This helps the model to perform well on data that it has never seen before and predict the output results. The opposite of supervised learning is unsupervised learning where the model tries to make sense of data itself without any guidelines and expected results from the data. Unsupervised learning models comprehend the data and try to detect similarities and patterns in the data, and categorizes it.</p><p>Supervised Learning can be broadly classified into two subcategories-</p><ol><li><strong>Regression</strong>- It is a kind of Supervised Learning algorithm where the algorithms learn from labeled datasets and it is able to predict a continuous-valued output for new data fed to the algorithm. Example of regression algorithms is- Predicting housing price, Predict Car price, Predicting Crime rates, etc. The major regression algorithms are- Linear Regression and Logistic Regression.</li><li><strong>Classification</strong>- Classification algorithms are kind of learning algorithms where they have to classify the new data into different classes. Unlike Regression, it doesn&#39;t have continuous-valued output. It is used where we need binary output, For example- Is the tumor malignant or benign, Does it will be a sunny day or not, Is a mail Spam or not, etc. The major Classifiers are- Decision Tree, Naive Bayes Classifier, and Support Vector Machines(SVM).</li></ol><p>Supervised learning is widely used in the practical world. The major practical applications of supervised learning are as follows-</p><ol><li>Spam Detection- This is used by Gmail, rediffmail, etc. It identifies words from the mail which can make it more likely to be spam mail. SO it classifies mails by identifying suspicious words in the mails. It keeps learning and improves with time.</li><li>Speech Recognition- It is an application we see in Alexa, Siri, Google Assistant, Cortana, etc. We teach the algorithm about our voice and how to interpret it. It keeps learning every time we use it and become more efficient.</li><li>Object Recognition- It is an application we see at various places. In this first, we train the algorithm on a large set of data, and then the algorithm recognizes different objects. It is largely used in Traffic systems, security systems, etc.</li><li>Bioinformatics- This is one of the most widely used applications of supervised learning algorithms, It is used in our smartphones for facial recognition, in-screen fingerprint, etc. It stores our biological information and uses it to recognize us and it gets better every time we use it.</li><li>Recommender Systems-It is used by popular sites like Netflix, Amazon Prime, Hotstar, etc. It recommends new movies for us based on our searches and previously watched movies.</li></ol><p>Now let&#39;s see how to implement Supervised Learning Algorithms by scikit learn with the help of an example problem. The example problem is “Predicting Life expectancy using Gapminder Dataset”.</p><p>But first, let&#39;s set some insights about scikit-learn. Scikit-learn is the most robust and useful library for machine learning in Python. It provides us with a variety of efficient tools for statistical modeling and machine learning. It is largely written in python and built upon other python libraries like NumPy, Matplotlib, and SciPy.</p><p>Let&#39;s Code for our problem statement i.e “<strong>Predicting Life expectancy using Gapminder Dataset</strong>”. It comes under the regression category, ok let&#39;s see how to use scikit-learn to implement the Supervised Learning model-</p><p>You can get information about the Gapminder dataset from here-<a href="https://www.gapminder.org/data/">Gapminder_Dataset</a>.</p><p>I will try to give an overview of the code without going deeper into technical definitions. So that it will help to get a holistic view of Scikit-Learn and Supervised Learning.</p><pre>import numpy as np</pre><pre>import pandas as pd</pre><pre># Read the CSV file into a DataFrame &#39;df&#39;</pre><pre>df = pd.read_csv(&#39;gapminder.csv&#39;)</pre><pre>y = df[&#39;life&#39;].values</pre><pre>X = df[&#39;fertility&#39;].values</pre><pre>y = y.reshape(-1,1)</pre><pre>X = X.reshape(-1,1)</pre><p>In this Code block, we have imported NumPy, pandas, and our gapminder dataset. After that, we have splitter our dataset into the target variable ‘y’ and features ‘X’. We then reshaped it so that it fits according to our needs.</p><pre>from sklearn.linear_model import LinearRegression</pre><pre># Create the regressor &#39;reg&#39;</pre><pre>reg = LinearRegression()</pre><p>Here we have imported LinearRegression from sklearn.linear_model module. And we have created an instance of it called ‘reg’.</p><pre># Create the prediction space<br>Here we have preloaded the fertility column in X to numpy array X_fertiity.</pre><pre>prediction_space = np.linspace(min(X_fertility), max(X_fertility)).reshape(-1,1)</pre><pre># Fit the model to the data</pre><pre>reg.fit(X_fertility,y)</pre><pre># Calculating the prediction</pre><pre>y_pred = reg.predict(prediction_space)</pre><pre># Print R^2 error</pre><pre>print(reg.score(X_fertility,y))</pre><pre># reg score is=0.6192442167740035</pre><pre># Plotting  the regression line</pre><pre>plt.plot(prediction_space, y_pred, color=&#39;black&#39;, linewidth=3)</pre><pre>plt.show()</pre><p>Here we have trained and tested our regressor ‘reg’ by using fertility as a feature and calculated accuracy and plotted the distribution of our fitted linear regression line.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TUzp0k8VqW07A6z5e5MJtA.png" /></figure><pre># Import necessary modules from sklearn</pre><pre>from sklearn.linear_model import LinearRegression</pre><pre>from sklearn.metrics import mean_squared_error</pre><pre>from sklearn.model_selection import train_test_split</pre><pre># Creating X_train, X_test, y_train, y_test</pre><pre>X_train, X_test, y_train, y_test = train_test_split(X,y, test_size = 0.3, random_state=42)</pre><pre># Create the regressor reg_a</pre><pre>reg_a = LinearRegression()</pre><pre># Fit the regressor to the training data</pre><pre>reg_a.fit(X_train,y_train)</pre><pre># Predict on the test data  &#39;y_pred&#39;</pre><pre>y_pred = reg_a.predict(X_test)</pre><pre># Compute and print R^2 and RMSE</pre><pre>print(&quot;R^2: {}&quot;.format(reg_a.score(X_test, y_test)))</pre><pre>#Prints , R^2: 0.838046873142936</pre><pre><br>rmse = np.sqrt(mean_squared_error(y_test,y_pred))</pre><pre><br>print(&quot;Root Mean Squared Error: {}&quot;.format(rmse))</pre><pre># Prints, Root Mean Squared Error: 3.2476010800377213</pre><p>In this Code block, we have trained our regressor reg_a with the whole X _train and y_train and tested on X_test and y_test. We had also printed the R² error and accuracy score of the model. We had used the sklearn module and imported LinearRegression, train_test_split, and mean_squared_error module which does all backend calculations. So this model can predict life expectancy now on new data with almost 84 percent accuracy.</p><p>So we can see that how easy it is to implement supervised learning models using Scikit-Learn. This is the power of Scikit-Learn which helps to easily implement any model without worrying about complex calculations.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=df617631216a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/analytics-vidhya/supervised-learning-with-scikit-learn-df617631216a">Supervised Learning with Scikit-learn</a> was originally published in <a href="https://medium.com/analytics-vidhya">Analytics Vidhya</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>