<?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[Leanplum Engineering - Medium]]></title>
        <description><![CDATA[Solving big challenges in data and engineering. To find out more about Leanplum’s integrated mobile marketing platform, visit www.leanplum.com - Medium]]></description>
        <link>https://medium.com/leanplum-engineering?source=rss----4f9e6a19d5a5---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Leanplum Engineering - Medium</title>
            <link>https://medium.com/leanplum-engineering?source=rss----4f9e6a19d5a5---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 16:50:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/leanplum-engineering" 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 front-end developer tools and workflows at Leanplum]]></title>
            <link>https://medium.com/leanplum-engineering/the-front-end-developer-tools-and-workflows-at-leanplum-826331e1cbda?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/826331e1cbda</guid>
            <category><![CDATA[storybook]]></category>
            <category><![CDATA[dependency-injection]]></category>
            <category><![CDATA[leanplum]]></category>
            <category><![CDATA[vue]]></category>
            <category><![CDATA[front-end-development]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Tue, 01 Jun 2021 12:46:22 GMT</pubDate>
            <atom:updated>2021-06-01T12:40:06.676Z</atom:updated>
            <content:encoded><![CDATA[<h3>The frontend developer tools and workflows at Leanplum</h3><h4>by Alex Gyoshev</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2xkuAV6TKPdO8sXRPdlG6Q.jpeg" /><figcaption>What do front-end developers do at Leanplum?</figcaption></figure><p>Before we focus on tools, let me give you a bit of context. What does Leanplum actually do? And what is the impact of front-end developers on the product?</p><p>Leanplum is a platform. It allows its customers to create marketing campaigns with push notifications, email and in-app messages. To set up audiences and A/B test message content and channels. To analyze the results, and create more engaging campaigns in the future. All of this is orchestrated in the Leanplum dashboard, a single-page application that we (as front-end developers) develop daily.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MJCjxzC6L9ORYWjY7d8lSQ.png" /><figcaption>Creating a campaign through the Leanplum dashboard.</figcaption></figure><p>In this post, I’ll share what our technology stack is, as well as some of the tools that boost our productivity.</p><h3>The Leanplum front-end stack</h3><p>To summarize, the front-end stack at Leanplum is based on Vue, with some influence by React and Angular. Over the years we’ve tried out different ideas, and the most sticky ones have made their way into our everyday workflow.</p><h4>The basics: Vue, class components, and TSX</h4><p>If you are familiar with Vue and single file components, you might be surprised by our setup. To illustrate, here is a sample component:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/136b3bcab3d0b72bd1b7d9be35378ffe/href">https://medium.com/media/136b3bcab3d0b72bd1b7d9be35378ffe/href</a></iframe><p>The <a href="https://github.com/kaorun343/vue-property-decorator">vue-property-decorator</a> library provides decorators that turn an ES6 class with properties into a Vue component (<strong>@Component</strong>) with props (<strong>@Prop</strong>). Notice that this is TypeScript, which provides type-checking and guards us from a lot of runtime errors.</p><p>This is also where TSX comes into play, which combines Vue render functions with the TypeScript type-checking, much like React.</p><h4>Code sharing: Custom Decorators and Dependency injection</h4><p>Next step: get ideas from Angular! Based on the above, we built our own set of tools that help with common use-cases in our daily work. These tools range from common utilities to services which provide an API, or decorators which automatically wire up Vuex. Combined with the <a href="https://inversify.io/">inversify</a> dependency injection library, we greatly reduce boilerplate.</p><p>One example that illustrates how we combine decorators and dependency injection is <strong>feature whitelists</strong>. Most of our non-trivial development is done behind feature flags, which allow us to build features incrementally and to have fine-grained control of their rollout.</p><p>Here is how a component can check if a customer is whitelisted for a feature and change its rendering:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/489aa618bbfdf82b0256c8627bd7d38c/href">https://medium.com/media/489aa618bbfdf82b0256c8627bd7d38c/href</a></iframe><p>The custom decorator <strong>@ContainerInject</strong> injects an implementation of the <strong>WhitelistProvider</strong> class, which is resolved by <strong>inversify</strong>. This makes it easy to test, and is easy to use.</p><p>After receiving sufficient feedback for the feature, we can flip the switch and deliver it to all customers (or decide to rework it). In the end we clean up the unused code branches.</p><p>A neat function of using dependency injection for the above is that we can provide a custom interface for admin users, without touching other code in the system. In particular, the following allows us to toggle whitelists through a click of a checkbox, so that we can quickly test features behind a whitelist:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*V0UyyX1rNWwz4LOoBH4vqw.png" /><figcaption>An interface that lets developers toggle whitelists for an application, without affecting its data.</figcaption></figure><h3>Reusable components and the Leanplum design system</h3><p>Another big boost in daily productivity is our component library, which implements the Leanplum design system. Apart from basic UI components like a Button or a NumberInput, this also includes bigger building blocks for the application, like page layouts or placeholder panes. These components make the dashboard UI more consistent and let us focus on more challenging problems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6H7L18aNBRXOBoSOd6S_Qw.png" /><figcaption>An interface that shows two UI patterns commonly seen in the dashboard: a single column layout (Title + top-right Button + Content area) and a placeholder in its center (Icon + Title + Subtitle + Button).</figcaption></figure><p>All components from the library are listed in <a href="https://storybook.js.org/">Storybook</a>, which shows usage scenarios and allows us to visually test components when making updates. This prevents us from breaking other usages by focusing too much on a specific issue and missing the big picture.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*A2HLcOYoeDABgzPNeHKJeQ.png" /><figcaption>Storybook lists all common components and shows configurations that are expected to work across the product.</figcaption></figure><h3>The path ahead</h3><p>It is easy to grow complacent with technology choices over time. To resolve this, our team holds regular front-end chapter meetings, where we discuss improvements that we could bring to our daily process. For example, we can decide when (and whether) we should migrate to Vue 3, or how we can improve asset bundling, or whether we need to change a time-handling library. This allows us to update or confirm our views, making sure that our tech stack is adequate for our daily work. It also fosters collaboration between all front-end engineers in the company, and breaks silos where we work too much on an isolated problem.</p><p>In the end, what makes a technology stack <em>good </em>is to be intuitive and to allow quick iterations. We’ve achieved that by getting good ideas from different frameworks, and by experimenting continuously.</p><blockquote><a href="https://jobs.lever.co/leanplum/059f4a0d-7cb1-4074-86ed-ef77d654100f?lever-via=9wnsqWLldd">Join our team</a> if you are passionate about the frontend tech stack we use at Leanplum!</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/256/1*MQEnAq6ycI-GA7JsmJi7Eg.jpeg" /></figure><h4>About the author</h4><p>Alex is a front-end developer, working on the Leanplum dashboard and the JavaScript SDK, among other things. He is fond of delivering good data-driven user experience and values robust tools.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=826331e1cbda" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/the-front-end-developer-tools-and-workflows-at-leanplum-826331e1cbda">The front-end developer tools and workflows at Leanplum</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Internship at Leanplum: Meet Viktor Terziev]]></title>
            <link>https://medium.com/leanplum-engineering/internship-at-leanplum-meet-viktor-terziev-bb7172bad5c6?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/bb7172bad5c6</guid>
            <category><![CDATA[backend]]></category>
            <category><![CDATA[intern]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[leanplum]]></category>
            <category><![CDATA[internships]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Tue, 13 Oct 2020 12:19:18 GMT</pubDate>
            <atom:updated>2020-10-13T12:28:50.628Z</atom:updated>
            <content:encoded><![CDATA[<p>Interns have always been an invaluable part of <a href="https://www.leanplum.com/">Leanplum</a> and critical to our company’s success! With access to a goldmine of data, Leanplum interns have the opportunity to innovate at scale and face some of the most challenging product development problems. Being part of the Leanplum Team means having ownership, direct impact, and opportunity to define the future of marketing. 🚀</p><p><em>We sat down with a few of our 2019 interns across multiple product development pods to hear their unique perspective about working at Leanplum (check out more stories </em><a href="https://eng.leanplum.com/"><em>here</em></a><em>).</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/472/1*EjNRaKUD4ypWpnXPhVL6Qw.jpeg" /></figure><h3>👋 Meet Viktor</h3><p>My name is Viktor and I am about to finish my second year at Sofia University “St. Kliment Ohridski”.</p><p>I found out about competitive programming in seventh grade and I have been enjoying it ever since. Another passion of mine since high school has been game development. 🎮 The thing is, however, that I have always been more interested in the challenges around making games, not actually playing them.<br>I consider that kind of programming as part of my spare time, but other things are also included in that category. I really like sports like basketball and volleyball.</p><p>I recently discovered climbing! We even went with the Leanplum team a couple of times. 🧗‍♂️</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_8eVC6ASwU4d2II54Ufzzw.jpeg" /></figure><h3>🔥 Why did you choose an internship at Leanplum?</h3><p>I actually heard about Leanplum from <a href="https://eng.leanplum.com/working-at-leanplum-the-intern-perspective-14c92ef7cf19">the previous interns</a>. We were hanging out and they just mentioned how amazing working here was.</p><p>One of them even took me to the office to show me around. After that I didn’t do any more research, just applied for an internship. And I still think it was a great choice!</p><h3>💻 What projects did you work on during your internship?</h3><p>I think one of the best things about my internship was that I didn’t have a separate, isolated project. I worked on whatever my team was working on.</p><p>That way you get much more feedback and learn much better from the senior engineers. I helped with a bunch of different things — some monitoring of our services, build optimization, etc.</p><p>My very first task was about the automatic resharding of one of our databases. After a while, I understood how important that database was and how many things could have gone wrong. But with the help of the team, I managed to do it and it’s still used in production, even today.</p><h3>🙌 What was your favorite part of the Leanplum internship experience?</h3><p>My favorite part was learning new things, which I guess is the main goal of every internship. I worked with so many different technologies, tools, frameworks. I had the opportunity to find out what I like and what I don’t.</p><p>Another amazing thing that I realized gradually is that the work you do at Leanplum really matters! Because the product is relatively small, you can actually make a noticeable impact right away. And that’s really motivating!</p><h3>🚀 What was your next career step after the internship?</h3><p>I really enjoyed my internship, so I wanted to continue working here. But I also wanted to continue studying at university, so I tried to combine the two.</p><p>As a result, I became a part-time employee and had the option of going to lectures and taking my exams. The working time is really agile. It’s not easy to do your studies and work at the same time, but I think I’m handling it well enough. 😉</p><p><strong>—</strong></p><p><strong>Leanplum is hiring! </strong><a href="https://jobs.lever.co/leanplum"><strong>Explore our career opportunities here</strong></a><strong>.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bb7172bad5c6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/internship-at-leanplum-meet-viktor-terziev-bb7172bad5c6">Internship at Leanplum: Meet Viktor Terziev</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Meet The Hiring Manager: Emil Stoychev, Leanplum ProdDev Lead]]></title>
            <link>https://medium.com/leanplum-engineering/meet-the-hiring-manager-emil-stoychev-leanplum-proddev-lead-a8c42db1de4c?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/a8c42db1de4c</guid>
            <category><![CDATA[hiring]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[engineering]]></category>
            <category><![CDATA[leanplum]]></category>
            <category><![CDATA[hiring-tips]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Mon, 17 Aug 2020 13:03:35 GMT</pubDate>
            <atom:updated>2020-08-17T13:03:35.008Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*9nzjZ8Xfr_cw-g1K.jpeg" /></figure><h3>👋 Hi, <a href="https://www.linkedin.com/in/emilstoychev/">Emo</a>! To start with, what does your role at Leanplum look like?</h3><p>As part of our Use Domain, my role involves ensuring the alignment between Leanplum’s priorities and our ProdDev teams (pods) as well as the importance of solving the most important product development challenges with the right people in place. As a Pod Lead of the API and the SDK pods, I’m accountable for all deliverables and KPIs of the team. I run the scrums and most important of all — I help people play to their strengths. At the end of the day, it all comes down to resolving roadblocks and driving continuous improvement!</p><p>The API pod, as its name suggests, owns the Leanplum API and aims to power the rest of the pods with industry-leading technology. Our customers and partners use it to provide us with information from their backend systems and integrate with Leanplum.</p><p>The SDK pod owns our iOS, Android, Unity, and React Native SDKs to allow mobile teams to integrate with our analytics, optimization, and messaging capabilities.</p><h3>Tell us more about your career journey 🛵</h3><p>In the past 15 years, I’ve been through different roles in the software industry and (as probably expected) made lots of mistakes along the way. :)</p><p>I started as an engineer in a small outsourcing company right after I enrolled in university. Fast forward a few years, I became a partner in the company and started dealing with clients and project management along with my engineering responsibilities. I was also in the technical training and consulting business for a few years — leading training courses around the world.</p><p>Then I stepped into product management and gathered experience both in the start-up (co-founding two companies) and the enterprise world (providing tools for software development). After about five years in the field, I found my way to Leanplum, where I initially joined as a product manager.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*RFMDmuVJ9B9JfKYX.jpeg" /></figure><h3>What about the Campaigns Pod at Leanplum? What does it stand for? 📲</h3><p>The Campaigns pod owns one of the core areas of our platform — the orchestration of marketing campaigns. This is what our customers use to connect with their users across multiple channels (mobile, in-app, web, and email).</p><p>A campaign can be a simple email sent to all users on a specific date, but can also be a complex multi-step, multi-channel campaign sent to a specific audience segment, based on their past and/or current behavior.</p><p>The challenges we solve range from quick data processing (~25K events per second only for campaign analytics), to efficient data storing (~25B data points per day), and managing infrastructure cost-efficiently.</p><p>The pod consists of front-end and back-end engineers, QA automation engineers, UX designers, and a product manager. It is a fully staffed pod that can execute on their own, considering both engineering and product perspectives.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/0*5VVD1V7UWhpF5hHn.jpeg" /></figure><h3>LEANPLUM is hiring 🚀 What interview tips would you give to anyone interested in joining the team?</h3><blockquote><em>Know your why!</em></blockquote><p>Why do you want to join our team? What do you stand for? How can you contribute to the success of Leanplum? What are your strengths and how do you want to continue your professional development? How do you expect Leanplum to support you in your career development?</p><blockquote><em>Come prepared and be ready for honest talk :)</em></blockquote><p>Both you and we need to be very confident in making the right choice so expect to have a round of interviews that look at you from different perspectives.</p><blockquote><em>Prepare your questions for us!</em></blockquote><p>What do you need to know to make sure we are the right fit for you?</p><h3>What are the most important skills/qualifications you are looking for?</h3><p>Skills vary for the different positions we are hiring for. Yet, something that is universal and important to us is:</p><ul><li>Are the challenges we are facing the kind of challenges you would like to solve? What is your background in this area?</li><li>How can your strengths map to the gaps we have in the pod?</li><li>How would you grow in three years from now?</li><li>How passionate are you to learn and solve difficult challenges?</li></ul><p>For the backend roles, we would evaluate additionally your expertise in cloud technologies, Kafka, Kubernetes, Spring Boot. We expect you to care both about infrastructure and backend.</p><h3>Why did you join Leanplum?</h3><p>Leanplum feels special! Because of the attention to detail in everything we do — from designing our working environment to building a world-class product, and hiring top professionals who define our company culture. You can simply be your unique self here!</p><p>During the interview process, I managed to learn a lot about the product and the team, thanks to the multiple steps involved and the different team members I met. The final step, a presentation in front of the whole team (a PM specific), gave me an opportunity to see what the collaboration and Leanplum team dynamics look like. And I loved it!</p><p>Another reason to join Leanplum was my interest in the field of marketing and big data. Given the scale of the product and the fact that it is an international company with offices in North America, Europe, and Asia, I knew it will bring many opportunities and different perspectives to learn from.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/0*YfPuGNUI05YMuSaR.jpeg" /></figure><h3>Interested to learn more about our Leanplum career opportunities? <a href="https://jobs.lever.co/leanplum">Explore How You Can Join Our Team here</a></h3><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a8c42db1de4c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/meet-the-hiring-manager-emil-stoychev-leanplum-proddev-lead-a8c42db1de4c">Meet The Hiring Manager: Emil Stoychev, Leanplum ProdDev Lead</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Internship at Leanplum: Meet Ivan Ganev]]></title>
            <link>https://medium.com/leanplum-engineering/internship-at-leanplum-meet-ivan-ganev-2cd797284911?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/2cd797284911</guid>
            <category><![CDATA[productdevelopmnet]]></category>
            <category><![CDATA[leanplum]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[internships]]></category>
            <category><![CDATA[computer-science]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Thu, 06 Aug 2020 11:04:06 GMT</pubDate>
            <atom:updated>2020-08-06T11:04:06.463Z</atom:updated>
            <content:encoded><![CDATA[<p>Interns have always been an invaluable part of <a href="https://www.leanplum.com/">Leanplum</a> and critical to our company’s success! With access to a goldmine of data, Leanplum interns have the opportunity to innovate at scale and face some of the most challenging product development problems. Being part of the Leanplum Team means having ownership, direct impact, and opportunity to define the future of marketing. 🚀</p><p><em>We sat down with a few of our 2019 interns across multiple product development pods to hear their unique perspective about working at Leanplum (check out more stories </em><a href="https://eng.leanplum.com/"><em>here</em></a><em>).</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*krou6OBxW3CduTFsk5hx7w.jpeg" /></figure><h3>👋 Meet Ivan Ganev</h3><p>I am <a href="https://www.linkedin.com/in/ivan-g-2b060410b/">Ivan</a> and I am currently in my second year of studying computer science at Sofia University “St. Kliment Ohridski”.</p><p>I love problem-solving! I feel lucky and grateful for being introduced to informatics at a very young age. The competitive nature and challenging programming problems got me hooked but the best part for me has always been the great friendships I made during these informatics competitions.</p><p>Another passion of mine is basketball 🏀 I love playing with friends, watching the NBA and video analytics.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GI5pgu2_vW3jjLpUdgEgRw.jpeg" /></figure><h3>🔥 Why did you choose an internship at Leanplum?</h3><p>Three of my very good friends interned at Leanplum the year before. They told me about the friendly environment, the smart colleagues, and the interesting projects they have been working on.</p><p>The idea of building a product that scales upon so many users appealed to me. Then I had a wonderful time during the interview process which sealed the deal and confirmed my expectations.</p><h3>💻 What projects did you work on during your internship?</h3><p>To get onboarded with the codebase, I was tasked initially with adding metrics to some of the Leanplum services. Then I built a Grafana dashboard for them. Finally, I had to set up the infrastructure that builds a data freshness report and sends it in a slack channel every week.</p><p>Next up was splitting one of the services into two smaller microservices. This task introduced me to many interesting technologies like Docker, Kubernetes, Spring Boot, and Kafka.</p><p>My latest task as an intern at Leanplum was to improve the accuracy of our campaign analytics. The focus was on a service that reads a message from Cloud Pub/Sub and transforms it into many records that are published to Kafka. Occasionally, the publishing would fail midway through and this would result in overcounting. Transactions solved the problem but were too slow so I implemented batching of the Pub/Sub messages. I also added deduplication of the messages because Pub/Sub often redelivered them.</p><h3>🙌 What was your favorite part of the Leanplum internship experience?</h3><p>My favorite part of Leanplum is the people!</p><p>I love the interesting discussions we have all the time and the fact that I can fully rely on my team in case I need some help.</p><p>Being part of Leanplum, I have learned many valuable lessons about life in general, not just about programming.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*W5xHicY2BR0V8fFlnHk0Dg.jpeg" /></figure><h3>🚀 What was your next career step after the internship?</h3><p>I had the opportunity to continue working as a full-time Software Engineer at Leanplum Use Domain while also continuing with my studies.</p><p><strong>—</strong></p><p><strong>Leanplum is hiring! </strong><a href="https://jobs.lever.co/leanplum"><strong>Explore our career opportunities here</strong></a><strong>.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2cd797284911" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/internship-at-leanplum-meet-ivan-ganev-2cd797284911">Internship at Leanplum: Meet Ivan Ganev</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Internship at Leanplum: Meet Denis Hristov]]></title>
            <link>https://medium.com/leanplum-engineering/internship-at-leanplum-meet-denis-hristov-b687258fb618?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/b687258fb618</guid>
            <category><![CDATA[internship-experience]]></category>
            <category><![CDATA[front-end-development]]></category>
            <category><![CDATA[employee-experience]]></category>
            <category><![CDATA[internships]]></category>
            <category><![CDATA[product-development]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Mon, 27 Jul 2020 19:05:49 GMT</pubDate>
            <atom:updated>2020-07-28T09:09:20.180Z</atom:updated>
            <content:encoded><![CDATA[<p>Interns have always been an invaluable part of <a href="https://www.leanplum.com/">Leanplum</a> and critical to our company’s success! With access to a goldmine of data, Leanplum interns have the opportunity to innovate at scale and face some of the most challenging product development problems. Being part of the Leanplum Team means having ownership, direct impact, and opportunity to define the future of marketing. 🚀</p><p><em>We sat down with a few of our 2019 interns across multiple product development pods to hear their unique perspective about working at Leanplum (check out more stories </em><a href="https://eng.leanplum.com/"><em>here</em></a><em>).</em></p><h4>👋 Meet Denis Hristov</h4><p>Hi, my name is <a href="https://www.linkedin.com/in/denis-hristov-660653175/">Denis Hristov</a>. I am a first-year student in computer science at Sofia University “St. Kliment Ohridski”.</p><p>I have been fascinated by the web ever since I discovered services like <a href="https://www.wix.com/">Wix</a> (before I started school) which help you create your own websites without even knowing what programming languages are. This sparked my creativity and developed into a passion for creating valuable and functional user interfaces.</p><p>Later on, during my high school years, I got my programming fundamentals and I fell in love with computer science. The dynamic and competitive environment there pushed me into putting my best and I eventually got my first internship.</p><p>Outside of programming, I enjoy doing some graphic designs. I recently moved into a new apartment so I am constantly checking out minimalistic interior concepts.</p><p>I joined Leanplum in the summer of 2019 right after graduating from Technology School Electronic Systems (TUES).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o8WlXZBvBbXVSfUGgxFRaQ.jpeg" /></figure><h4>🔥 Why did you choose an internship at Leanplum?</h4><p>A good friend of mine recommended the company and I was immediately interested to learn more.</p><p>I attended a few <a href="https://www.meetup.com/Leanplum-Tech-Talks-Sofia/">Leanplum Meetups</a> where I got a taste of the company’s culture and product. Both resonated a lot with me! Especially the product part since I love Leanplum’s vision and wanted to put my commitment and passion for building a valuable product into something good.</p><h4>💻 What projects did you work on during your internship?</h4><p>I primarily contributed to delivering market-leading Audience Management experience which was a brand new part of the product. The goal is to help our clients better understand their audiences and have confidence and trust when targeting them with marketing campaigns. I took an active part in developing the UI components that would build the new experience, helped design how it would integrate into the already existing parts of the product and voiced my views on key product and design decisions.</p><p>At Leanplum we have our own outstanding design system which includes a collection of reusable UI components that we build and maintain. I loved collaborating with designers on implementing some of them and later using them to build the product’s interface.</p><p>One of the more challenging tasks I worked on during the internship was implementing the undo/redo functionality of the Audience Builder. Existing solutions did not meet our requirements so we had to settle for a custom one. It also had to support Vue’s reactivity which required reading its source code and understanding how it works so we can be compatible. The trivial solution was to keep a list of each of the audience’s states and switch between them but it had the downside of allocating too much memory, most of which contained the same content. So we had to put some reasonable constraints.</p><p>I researched how modern libraries solve similar issues like tracking changes to a nested object and reacting to them. I got inspired to solve the problem similarly using JavaScript’s Proxy. It allowed me to define custom handlers for fundamental operations like setting values and defining properties to an object while also having the upside of not requiring me to deep copy each state. The solution came in neatly and was quite a lot less memory consuming (but also a little more complex 😁).</p><h4>🤔 What is it like to be an intern at Leanplum?</h4><p>Honestly, the experience was one of a kind! Everyone was quite welcoming and joyful. Getting to know the company’s culture was the best part since I resonated a lot with it.</p><p>The cool thing was that it did not really feel like an internship since I was solving real engineering problems and my opinions and views were respected by the team.</p><h4>🙌 What was your favorite part of the Leanplum internship experience?</h4><p>Definitely the team! I have met some truly inspiring and visionary people who have motivated me throughout the internship as well as many talented and skilled individuals from whom I’m still learning a lot. Everyone is quite friendly, supportive, and a good company (even outside working hours)!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o91Bg3WBYRz9RXYSi8vA-g.jpeg" /></figure><h4>🚀 What was your next career step after the internship?</h4><p>I did transition into a full-time employee at Leanplum Audience Insights Pod and even got promoted to Eng II recently.</p><p>I love working here and I am genuinely excited about what the future holds for Leanplum.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Th3Ogax9_q6eGn0dJMeP2g.jpeg" /></figure><p><strong>—</strong></p><p><strong>Leanplum is hiring! </strong><a href="https://jobs.lever.co/leanplum"><strong>Explore our career opportunities here</strong></a><strong>.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b687258fb618" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/internship-at-leanplum-meet-denis-hristov-b687258fb618">Internship at Leanplum: Meet Denis Hristov</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How do we deploy front-end code asynchronously: Vue Microfrontends]]></title>
            <link>https://medium.com/leanplum-engineering/how-we-deploy-front-end-code-asynchronously-vue-microfrontends-2947384292c9?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/2947384292c9</guid>
            <category><![CDATA[leanplum]]></category>
            <category><![CDATA[front-end-development]]></category>
            <category><![CDATA[frontend]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[vue]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Wed, 01 Jul 2020 08:32:42 GMT</pubDate>
            <atom:updated>2020-07-01T08:34:23.103Z</atom:updated>
            <content:encoded><![CDATA[<h4>by Alex Gyoshev</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DZe2_Kj4G9M0Pw2H-str7g.jpeg" /></figure><h3>The Monolith</h3><p>One essential part of the <a href="https://www.leanplum.com/">Leanplum</a> platform is the Dashboard website. It is the part of Leanplum that our customers use daily. It is where they create multi-channel marketing campaigns, slice and dice target audiences, and create email templates.</p><p>The front-end code of the Leanplum dashboard is part of a weekly monolith release, along with all required services. This is a good pace for mature parts of the project and allows a thorough QA process.</p><h3>The Problem</h3><p>For areas in active development, though, the weekly release pace is too slow. During the closed beta, for example, we strive to address our customers’ feedback in a matter of days, if not hours. And updating all front-end assets due to a change in a beta service is risky. For instance, we want to innovate and ship our new audiences dashboard quickly without affecting campaign creation.</p><p>Because of this, we looked for ways to make changes to the new components fast and isolated, while maintaining the monolith release for the whole system.</p><h3>Types of Microfrontends</h3><p>Microfrontends is an umbrella term for different ways to break apart a big single-page application.</p><p>Here’s a taste of the different flavors.</p><p>On one end of the spectrum, there is the <strong>complete isolation between components</strong>. This allows different teams to deploy parts of the application with any technology they choose. For example, one team may use Vue, and another may use React. The output of each is a component that is rendered by a lightweight router, with no code shared between components. While this approach allows complete freedom within an autonomous team, it also creates silos in the front-end organization, because it makes cross-team PR reviews very hard. It’s hard to work effectively with multiple stacks at once.</p><p>On the other end of the spectrum is to keep a single technology stack and make <strong>standalone deployments for isolated components</strong>. While this is a powerful approach, it holds the problem that components are very constrained to work with the surrounding page. Any changes in routing need to be handled by redeploying the whole application.</p><p>In the end, our approach ended somewhere in the middle. But before we got there, we had to list the challenges we wanted to be fixed.</p><h3>Challenges</h3><p>Before trying out different approaches, we defined a set of constraints to check the feasibility of each potential solution:</p><ul><li>The solution should allow the gradual migration of parts of the dashboard to Microfrontends.</li><li>Our common UI library (aptly named lib-ui) should be updated only with the monolith.</li><li>Microfrontends should come with a trivial cost to the bundle size.</li><li>We should be able to revert Microfrontend deployments, if necessary.</li></ul><p>All these requirements need to be answered by the approach that we tried.</p><p>Also, we had to answer a very important question about how to connect Microfrontends in the application. For example, here are some tasks that have to work with different Microfrontends:</p><ul><li>Build a target audience (audiences module) and create a new campaign for it (campaigns module).</li><li>Create a new audience (audiences module) from the users that have reacted positively to a specific campaign message (campaigns module).</li><li>Editing a message content (content module) while building a campaign (campaigns module).</li></ul><h3>Our Solution</h3><p>We decided on a common set of modules that are shipped as part of the monolith release. This includes common external dependencies like vue and vue-router, as well as the UI library (lib-ui) and common utilities. We call this the application shell.</p><p>To give full control on a Microfrontend, we bundle its routes and components into a single module that can be deployed on its own.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/8bc8ab4ca8d3e8818cd70e9e4a9ec930/href">https://medium.com/media/8bc8ab4ca8d3e8818cd70e9e4a9ec930/href</a></iframe><p>Every deployment record contains audit information, making it easy to trace and revert if necessary.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F-K7x8SvjWtbABpG6uiQUQ.png" /></figure><p>To reduce dependencies between the modules, we agreed on routing contracts or to share data through the <strong>vuex</strong> store. For example, instead of creating a new <strong>Campaign</strong> entity in the <strong>audience</strong> module, we call a route that is handled by the <strong>campaigns</strong> module. This severs the direct dependency and replaces it with a contract.</p><pre>Create Campaign (audiences module)<br>-&gt; /campaigns/new?audience=my-audience (common contract in router)<br>-&gt; New Campaign (campaigns module)</pre><p>This allows us to ship both the audiences and campaign modules separately if we ensure that the contract is followed.</p><h3>In Summary</h3><p>We like this mixed approach because we value the stability that comes from the cadence of our monolith releases, but we also see those quick front-end deployments are invaluable when we innovate on new parts of the product. After introducing it, we have gradually migrated several parts of our front-end application to this approach. We even managed two in a day, during a recent odd week (a week at the beginning of each quarter that encourages creativity and innovation by providing Leanplum members the opportunity to work on new ideas)!</p><p>The approach proved handy on several occasions already, allowing us to deliver UI fixes without redeploying services. With this in place, we started shifting our thinking towards making parts of the application smaller and independent, which in turn makes them more maintainable.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2947384292c9" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/how-we-deploy-front-end-code-asynchronously-vue-microfrontends-2947384292c9">How do we deploy front-end code asynchronously: Vue Microfrontends</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Effectively Replace explode() with flatMap() in Spark Dataset]]></title>
            <link>https://medium.com/leanplum-engineering/how-to-effectively-replace-explode-with-flatmap-in-spark-dataset-66b333d372ce?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/66b333d372ce</guid>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[spark]]></category>
            <category><![CDATA[product]]></category>
            <category><![CDATA[productdevelopmnet]]></category>
            <category><![CDATA[leanplum]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Wed, 13 May 2020 07:06:01 GMT</pubDate>
            <atom:updated>2020-05-13T08:03:15.733Z</atom:updated>
            <content:encoded><![CDATA[<h4>by Diman Karagiozov</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/958/1*krmd6O9rPKelp7SWI6oSUA.jpeg" /></figure><p>When it comes to working with big data, one often has to shift from classic relational designs to stream processing patterns.</p><p>Let’s see how our team at Leanplum solved some significant business, cost, and engineering challenges by changing the paradigm.</p><h3><strong>Problem statement</strong></h3><p>Imagine you are charging on API calls, depending on the payload. For example, <em>/setUserAttributes </em>sets one or more user attributes and costs 1¢, <em>/track</em> adds specific event information to a single user record and again costs 1¢. Moreover, different API calls can be <strong>batched</strong> (e.g. a single API call may consist of <em>track, track, setUserAttributes, track</em>).</p><p>Your task is to provide, for a given period of time (e.g. month), an aggregated view of billing information attributed to each request your API methods receive.</p><p>Going more on the technical side, each API method may or may not contribute to the billing with one or more “contributions”. Each contribution may have one or several “values” (details). The required aggregated information depends on the day the request happened, the number of contributions, and specific business logic based on the nature and relation between contributions and values.</p><p>Without being too specific an Entity Relationship diagram mapping to the data model might look like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*3f7mIMNFRaySykI9" /></figure><p>SQL query providing the needed aggregation would definitely do two JOINs to “flatten” the billing record, contribution, and details data. Then specific filtering would be applied (e.g. by time frame) and the specific business logic would be implemented as IF / THEN (or CASE / WHEN) statements (depending on the SQL database in use). Finally, there would be several HAVING and GROUP BY clauses.</p><p>A schematic query may look like this:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/bc6032d9cf7125df56017488280905c3/href">https://medium.com/media/bc6032d9cf7125df56017488280905c3/href</a></iframe><p>This would be a hassle-free solution if your API is low QPS (queries per second). When it comes to billion requests per day, no relational database could run the data ingestion (inserts) or the aggregation queries in a reasonable time.</p><p>The scenario described above is not that imaginary :) Here at Leanplum we have 10 billion API requests per day and we do provide our customers with aggregated billing data on a daily basis.</p><h3><strong>So how do we do it</strong></h3><p>The data ingestion process (which will be a topic of another blog post) is based on Apache Beam where each API request is transformed into a BillingRecord. The billing records stream is ingested in parquet files with the following schema (the schema is simplified for the purposes of the example):</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/9ed060439229472af772624ff299d2b1/href">https://medium.com/media/9ed060439229472af772624ff299d2b1/href</a></iframe><p>Then the querying is done within a Spark cluster and the aggregated results are again saved in parquet files (with simpler and flat schema).</p><h3><strong>Hands-on the data</strong></h3><p>Working with parquet files and Spark requires a lot more preparation work than just running a SQL query. So we first loaded the billing data for a month and customers with Zeppelin and worked on a SQL-like query. Not only we wanted to be more familiar with the data, but also to test how well such a solution would scale. Not surprisingly the query had two LATERAL VIEW statements (alternative of JOINS in SQL) and more or less the same HAVING and GROUP BYs.</p><p>The performance was poor (more than 1 hour for data for one customer for one day), even with fully scaled out Spark cluster (e.g. a hundred workers with 8 CPUs each and 32 Gb of RAM). So basically this was a NO-go approach.</p><h3><strong>explode() solution</strong></h3><p>A key takeaway from the previous attempt was that our data is really BIG — it is impractical and impossible to aggregate all data for all customers for a whole month. Thus, further on, we’ll be aggregating data for all customers but for a single day.</p><p>Our next attempt was to create a tailor-made Spark job and deploy it in the cluster. The full source code of the <strong>ExplodeJob</strong> can be found in the attached source code archive.</p><p>As you can see, there is one user-defined function (UDF), doing intermediate aggregation (lines 46–51) and two “explode()” statements used for surfacing the data needed for the specific business logic (lines 56–58).</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/1237052f37c703194090936023880ce1/href">https://medium.com/media/1237052f37c703194090936023880ce1/href</a></iframe><p>The method <strong>customFilteringLogic</strong> (lines 80–93) and <strong>operationsContrib</strong> (lines 68–71 and 95–103) implements the specific business logic (e.g. not all needed to make the required aggregation) and the final aggregation through a cumbersome Spark Dataset DSL.</p><h4><strong>So, the big question is: What is the performance of this solution?</strong></h4><p>Well, on <strong>average 2 hours</strong> for each day of the month, with <strong>spikes up to 4 hours</strong> for really busy days. These results are acceptable from a business point of view and we are able to deliver the requested aggregation for all customers on a daily basis.</p><p>On the downside, however:</p><ul><li>This solution is quite pricey, from a cost perspective;</li><li>From an engineering perspective, 2 hours average with spikes to 4 hours is a huge opportunity for optimization;</li><li>From a testing perspective, this approach is extremely hard to test. How would you write a unit test for the specific filtering and the business logic in the aggregation functions?</li></ul><h3><strong>flatMap() solution</strong></h3><p>At Leanplum we have never been satisfied with a solution that is “just OK” or “good enough”. So naturally we started looking for a better solution.</p><p>Analysis of the Spark job execution plan showed that most of the time and resources are spent in exploding the two nested layers of data. Then filtering would have cut 80% of the intermediate data and finally the aggregation would have reduced the output rows to a fraction of a percentage of the fully exploded data.</p><h4>So the next big question was: <strong>How to optimize the two explode() calls and even bolder — Do we actually need the explosions at all?</strong></h4><p>To find the answer, we needed a mind shift — from SQL (relational) thinking towards stream processing!</p><p>It was crucial to understanding that a billing record is independent of the rest of the billing records. So, basically, all the required information has been already loaded in the billing record and we only needed to transform it. Naturally, “flatMap()” kicked in. The full source code of the <strong>FlatmapJob</strong> can be found in the attached source code archive.</p><p>The structure of the job is greatly simplified:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/eed0068672792dad50a6cd5f002f5391/href">https://medium.com/media/eed0068672792dad50a6cd5f002f5391/href</a></iframe><p>The heavy business logic is implemented in a separate Java class — <strong>com.leanplum.cases.spark.jobs.ValuesFlatMapFunction</strong></p><p>The semantics of flatMap is simple — one object comes in, zero, one or more objects come out of the flatMap transformation. In the case above, the input is a single Dataset Row and the output is a <em>collection of Rows</em>, however, with an intermediate schema (provided by <strong>flattenedSchema</strong>) which is then directly used for the final aggregation.</p><p>The evaluation of the performance of flatMap() solution showed around <strong>40 minutes average</strong> (per day) processing time with <strong>spikes up to 1 hour and 20 minutes</strong> for requests-heavy days.</p><p>Besides the obvious performance gain the flatMap () solution has another great benefit — we are now able <strong>to cover the whole business logic with fast unit tests</strong> (on the contrary, to test the explode() solution we had to make heavy integration tests with local Spark instances).</p><h3><strong>Conclusion</strong></h3><p>flatMap() turned out to be the sole solution to our business, cost, and engineering challenges.</p><p>To effectively work with big data, engineers have to make a mind shift. In our case, this meant transitioning from thinking in SQL to thinking in Streams.</p><p>Being far from thinking explode() vs. flatMap() is the only way to optimize data processing, I have a question for you.</p><h4><em>What else could be optimized in the scenario described above?</em></h4><p>Let’s discuss alternative ideas and approaches!</p><p>Find the source code <a href="https://github.com/diman-karagiozov/leanplum/tree/master/SparkFlatMap">here</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=66b333d372ce" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/how-to-effectively-replace-explode-with-flatmap-in-spark-dataset-66b333d372ce">How to Effectively Replace explode() with flatMap() in Spark Dataset</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Product Development Spotlight: Meet Silvia Bakalova, Data Scientist at Leanplum]]></title>
            <link>https://medium.com/leanplum-engineering/product-development-spotlight-meet-silvia-bakalova-data-scientist-at-leanplum-a9b414d1b0b0?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/a9b414d1b0b0</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[leanplum]]></category>
            <category><![CDATA[software]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Wed, 06 Mar 2019 06:01:00 GMT</pubDate>
            <atom:updated>2019-03-06T06:01:00.828Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*03VZktCdEtA379XW" /></figure><h3>Hi, Silvia! To start, what is your role at Leanplum?</h3><p>I am a Data Scientist at Leanplum Omni-Channel Insights pod. Our pod delivers flexible and scalable analytics on a huge amount of users data coming from different channels. We are also responsible for data ingestion of more than 20 billion events per day, as well as storage and processing of that data that is close to four petabytes now.</p><p>My role as a Data Scientist is to support strategic and product initiatives. Leanplum is every inch a data-driven company and we use a lot of metrics to track, optimize, and improve our service and performance. I create statistical models and dashboards.</p><p>As we have huge amounts of data, we also utilize machine learning algorithms to optimize resources and enhance user experience. As a data scientist, my focus is also on double checking the data to ensure the accuracy of the insights we provide to customers.</p><h3><strong>What is your background in data science and what led you to Leanplum?</strong></h3><p>My background in science initially started in physics as I have obtained a Ph.D. degree in Physics. The transition from modeling data from physical experiments towards creating other predictive models from other sources of information came as a natural step for me.</p><p>Also, I have always been passionate about computers and AI. Artificial intelligence was the essence of my previous company, <a href="http://connecto.ai/">Connecto.ai</a>, where the team and I worked on creating natural language conversational systems. In 2018, Connecto was acquired by Leanplum, blending together technologies, data, and engineers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*yE63yyxS4pKvrhLu" /></figure><h3>From the perspective of a data scientist, what makes Leanplum the company of choice for top product development talents?</h3><p>Leanplum started as an A/B testing platform for mobile applications and as such, I think everything is done in a clever way. From organization, planning, development, and support — all aspects are modern and live and improving all the time. People test their ideas, so changes are fast and safe.</p><p>Leanplum is a rare type of software company with a big customer base, a complex product, and a safe, supportive environment where vision and ideas are shared and tested. That makes everyone involved — and every contribution is visible and valued!</p><h3>What excites you about working with data?</h3><p>When you are working with data, there is always something unknown and unexpected. Even when you are just confirming a hypothesis, there is a real joy that things are the way you expect. It is like solving puzzles — asking questions all the time and finding the answers in the data. Statistics can prove you right or prove you wrong!</p><p>With machine learning, you have an even more powerful way to test ideas and work with data in a dynamic way.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*KNiN_Y0ZFhwZlaJ8" /></figure><h3>What’s been your favorite challenge to work on at Leanplum?</h3><p>There are many trade-offs when you work with big data on the scale that Leanplum does. You need to balance between performance, flexibility, and cost. At the same time, you have to prepare for the next level of scale. What happens if the data doubles? What happens if all clients start using data-intensive processing? Part of those challenges could be solved with machine learning. I am glad to have the opportunity to work on those challenges.</p><h3><strong>What is Leanplum culture like through your point of view?</strong></h3><p>I have a favorite quote by Antoine de Saint-Exupéry that I would say greatly describes the culture at Leanplum:</p><blockquote>“If you want to build a ship, don’t drum up the men to gather wood, divide the work, and give orders. Instead, teach them to yearn for the vast and endless sea.”</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*1OSTpq71zVDreylc" /></figure><h3>Which books/resources would you recommend for someone looking to get into data science?</h3><p>Now there are plenty of resources to learn data processing, statistics, and machine learning. Working with large volumes of data is also now more accessible to cloud technologies.</p><p>For me, an interesting book that changed my views about programming was <a href="https://www.amazon.com/Bayesian-Methods-Hackers-Probabilistic-Addison-Wesley/dp/0133902838"><em>Bayesian Methods for Hackers: Probabilistic Programming and Bayesian Inference</em></a> by Cameron Davidson-Pilon. This book demonstrates with examples of how computers can “start thinking probabilistic” — like humans do.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/895/0*CjlN747Q1G50vY_F" /></figure><h3>What would you say to someone considering joining Leanplum?</h3><p>Leanplum is a great place to grow professionally and as the company culture is defined — to build meaningful relations.</p><p><strong>Leanplum is hiring! </strong><a href="https://www.leanplum.com/company/careers/"><strong>Explore our career opportunities here</strong></a><strong>.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a9b414d1b0b0" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/product-development-spotlight-meet-silvia-bakalova-data-scientist-at-leanplum-a9b414d1b0b0">Product Development Spotlight: Meet Silvia Bakalova, Data Scientist at Leanplum</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Product Development Spotlight: Meet Lev Neiman, Senior Software Engineer at Leanplum]]></title>
            <link>https://medium.com/leanplum-engineering/product-development-spotlight-meet-lev-neiman-senior-software-engineer-at-leanplum-8d2f5c63643d?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/8d2f5c63643d</guid>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[engineering]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[leanplum]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Thu, 21 Feb 2019 06:01:02 GMT</pubDate>
            <atom:updated>2019-04-23T11:52:52.941Z</atom:updated>
            <content:encoded><![CDATA[<h3>Product Development Spotlight: Meet <strong>Lev Neiman</strong>, Staff Software Engineer at Leanplum</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*rY01SA_qBkH5FUth" /></figure><h3>Hi, Lev! To start, what is your role at Leanplum?</h3><p>I started June 2017 on Leanplum Orchestration Pod and helped with the development of our new Campaign Composer dashboard — in particular, the live previews functionality.</p><p>I then moved on to working directly with our co-founder Andrew First on the Costs and Reliability team. I have done a lot of work around data consistency, scalability, reliability, and efficiency.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*O5nHqBEwnT0pAmrY" /></figure><h3><strong>What is your background in engineering and what led you to Leanplum?</strong></h3><p>I was tremendously lucky by having awesome parents who encouraged my curiosity. My father got me a QBasic book when I was around 10 and I fell in love with programming. When the time came to pick a major for college, Computer Science was a natural choice for me.</p><p>I have worked at big companies like <a href="https://www.amazon.com/">Amazon</a>, <a href="https://www.linkedin.com/">LinkedIn</a>, and <a href="https://www.reddit.com/explore">Reddit</a> where I have done pretty much everything — from front-end and mobile development all the way to the nitty-gritty of back-end distributed systems. However, things always moved a bit slower than I would have liked, so I decided to join a smaller company. Leanplum reached out to me at just the right time!</p><p>I had a great interviewing experience and afterwords Momchil, our CEO, took me for a walk-and-talk around the neighborhood. His charismatic history lesson of Leanplum finally led me to become part of Leanplum team.</p><h3>From the perspective of an engineer, what makes Leanplum the company of choice for top IT talents?</h3><p>Joining Leanplum, I got my wish for speed! Things can move incredibly fast here, and there is a lot of opportunity for ownership of existing and new code. A scale of traffic compared to the number of engineers is great!</p><p>Sometimes things can get intense, but you always have an incredible and supportive group of very talented engineers to help you out, brainstorm, or just socialize with.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*NKomWmd0TYJHT9SQ" /></figure><h3>What’s been your favorite challenge to work on at Leanplum?</h3><p>Maintaining, optimizing, and re-architecting content assignment architecture.</p><p>It is a critical piece of our business which assigns and renders content, such as emails, push notifications or in-app messages to our customer’s end-users. Content assignment gets invoked up to 27k/second on our public API, and every time we send messages (almost 1 billion times a day!)</p><p>So far I managed to save over 20 percent in CPU costs, reduce latency, and fix and prevent serious data consistency issues at high scale.</p><h3><strong>What do you most look forward to every day?</strong></h3><p>I love writing and reviewing code. Brainstorming with my peers is amazing as well, as we all learn something new. There are so many new things to consider and analyze!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*7BBoV1_dNMJt1kvy" /></figure><h3>What would you say to someone considering joining Leanplum?</h3><p>If you want to challenge yourself and grow, join us and help us solve unique problems at scale.</p><p><strong>Leanplum is hiring! </strong><a href="https://www.leanplum.com/company/careers/"><strong>Explore our career opportunities here</strong></a><strong>.</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8d2f5c63643d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/product-development-spotlight-meet-lev-neiman-senior-software-engineer-at-leanplum-8d2f5c63643d">Product Development Spotlight: Meet Lev Neiman, Senior Software Engineer at Leanplum</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Product Development Spotlight: Meet Petar Petrov, Principal Software Engineer]]></title>
            <link>https://medium.com/leanplum-engineering/product-development-spotlight-meet-petar-petrov-principal-software-engineer-3e86eafbb17b?source=rss----4f9e6a19d5a5---4</link>
            <guid isPermaLink="false">https://medium.com/p/3e86eafbb17b</guid>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[leanplum]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[engineering]]></category>
            <dc:creator><![CDATA[Leanplum Engineering]]></dc:creator>
            <pubDate>Thu, 17 Jan 2019 10:32:56 GMT</pubDate>
            <atom:updated>2019-01-17T10:31:45.278Z</atom:updated>
            <content:encoded><![CDATA[<h3>Hi, Petar! To start, what is your role at Leanplum?</h3><p>My role as a Principal Software Engineer and Back-End Chapter Lead at Leanplum is to drive alignment on best practices and patterns for building our back-end. This includes also testing these practices myself to make sure they are reasonable.</p><p>As I like to say, my role is to make sure engineering is pleasant for the engineers and actually creates value for the customers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*s4hkLYHgvI6fJ7Ox" /></figure><h3>What does your day-to-day look like?</h3><p>Most of my day is spent on engineering and meetings with different stakeholders. I also find time to study quality open-source projects and read books.</p><p>My engineering work is focused on improving the established practices for building products within the back-end chapter. This mostly includes:</p><ul><li>Spotting common engineering/technical problems</li><li>Finding the right people who can contribute to defining a canonical way of solving those problems</li><li>Driving consensus on the proposed solution</li><li>Announcing, documenting and explaining the proposed solution</li><li>Challenging and testing the decision by actually applying it as a real case situation</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*8t3H2HE6qABQyYdt" /></figure><h3><strong>What is a Chapter in terms of Leanplum organization structure?</strong></h3><p>Chapters (along with Pods) are key elements of Leanplum’s Product Development organization structure. Inspired by best practices from companies like Spotify and Amazon, we build autonomous, full-stack teams and empower them to innovate at scale.</p><p>Chapters are important because they drive alignment around common practices and technology choices and enable Leanplum teams (pods) to be successful in chapter functional areas like FE, BE, Mobile, Design, and PM. Chapters aim to initiate projects that benefit teams and provide tech expertise, coaching, mentorship, blueprints, and growth opportunities.</p><h3>What is your background in engineering and what led you to Leanplum?</h3><p>I believe that great things are built only when passion and self-improvement are one’s main motivation. Engineering has been my passion since I was 13 years old.</p><p>My background is quite diverse. I have built many things in the past and am still enjoying doing that. The part of my background, however, that I believe made the most impact in the engineering community, was the creation of Protocol Buffers, version two, with the goal of open-sourcing it. We were three engineers, working on this in our 20 percent time, getting help and tons of critics from Jeff Dean and Sanjay Ghemawat back in 2008 when I was working at Google. The project had a fairly low priority for the company and we worked on it simply because we thought it was cool.</p><p>Soon after we released it, however, more and better serialization formats came up. We helped the open-source community make a step forward and today PB v2 is used in so many open-source projects — far beyond what we anticipated.</p><p>What led me to Leanplum is Momchil Kyurkchiev — a passionate co-founder with a bold idea to transform a whole industry. The opportunity to optimize marketing while solving the colorful pile of technical challenges the business is facing was (and still is) quite appealing to me.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*NNAXSRkURsjM7ICA" /></figure><h3>From the perspective of an engineer, what makes Leanplum the company of choice for top IT talents?</h3><p>For me, it’s the fact that the company is alive — dynamic, prosperous, full of potential! I believe the company is alive because people are alive — curious, talented, and passionate professionals sharing common values.</p><p>And people are alive because there are multiple technical (and organizational) challenges to tackle on a massive scale of innovation and impact for global brands.</p><p>I love working on that while interacting with truly smart and interesting people.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/0*PrNfLS7CfLdMhEgO" /></figure><h3>What’s been your favorite challenge to work on at Leanplum?</h3><p>The new generation of analytics product. Collecting and applying learnings of data infrastructure, processing, storage, and retrieval has been so far my favorite technical challenge.</p><p>The non-technical challenges are also quite exciting. At the end of the day, it’s important to make sure that engineering is both being fun for team members and creating value for customers.</p><h3>What opportunities does Leanplum have for product development specialists?</h3><p>The opportunities are numerous. The most obvious ones are in the areas of Big Data (infrastructure, storage, processing, etc.), DevOps, and Machine Learning.</p><p>The impact of Leanplum current projects can be found in different areas like scalability, reliability, resiliency, automation of various operations, and last but not least, the optimizations of our products.</p><p>I strongly believe that the best opportunities are those that an engineer (or a PM, PL, etc.) recognizes to be the right ones based on their own passion, skill set, and love.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*MMHLbggR5UbUWhW4" /></figure><h3>Reading is an essential part of your day. Which books/resources would you recommend?</h3><p>There are many excellent technical books with a different focus. But the ones I would recommend are the following books by Kerry Patterson and Joseph Grenny:</p><ul><li><a href="https://www.amazon.com/Crucial-Conversations-Talking-Stakes-Second/dp/0071771328/ref=sr_1_1?ie=UTF8&amp;qid=1546803504&amp;sr=8-1&amp;keywords=Kerry+Patterson">Crucial Conversations</a></li><li><a href="https://www.amazon.com/dp/0071808868?aaxitk=1S4R1oU9oUIEAGyLOqfn9g&amp;pd_rd_i=0071808868&amp;pf_rd_p=3ff6092e-8451-438b-8278-7e94064b4d42&amp;hsa_cr_id=6205478540201&amp;sb-ci-n=asinImage&amp;sb-ci-v=https%3A%2F%2Fimages-na.ssl-images-amazon.com%2Fimages%2FI%2F61N3Fr1U%2BJL.jpg&amp;sb-ci-a=0071808868">Influencer</a></li><li><a href="https://www.amazon.com/Crucial-Confrontations-Resolving-Promises-Expectations/dp/0071446524/ref=sr_1_1?s=books&amp;ie=UTF8&amp;qid=1546803668&amp;sr=1-1&amp;keywords=crucial+confrontations">Crucial Confrontations</a></li></ul><p>They are not much about engineering, but about skills that can significantly improve one’s relationships (both personal and professional).</p><p>As for engineering, there is the old and timeless book by Andrew Hunt, <a href="https://www.amazon.com/Pragmatic-Programmer-Journeyman-Master/dp/020161622X/ref=sr_1_1?s=books&amp;ie=UTF8&amp;qid=1546803826&amp;sr=1-1&amp;keywords=the+pragmatic+programmer"><em>The Pragmatic Programmer</em></a>, and many more that are excellent.</p><p>Just read them all — in order of personal preference!</p><h3>What would you say to someone joining Leanplum?</h3><p>Take your time to find the challenge that inspires you and let’s start working.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*TO2Jx60JDYcqCILH" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3e86eafbb17b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/leanplum-engineering/product-development-spotlight-meet-petar-petrov-principal-software-engineer-3e86eafbb17b">Product Development Spotlight: Meet Petar Petrov, Principal Software Engineer</a> was originally published in <a href="https://medium.com/leanplum-engineering">Leanplum Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>