<?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 Vitalii Duk on Medium]]></title>
        <description><![CDATA[Stories by Vitalii Duk on Medium]]></description>
        <link>https://medium.com/@vitd?source=rss-f48e1b30e45------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*QVRHeaArFT92CxQ0rmo40Q.jpeg</url>
            <title>Stories by Vitalii Duk on Medium</title>
            <link>https://medium.com/@vitd?source=rss-f48e1b30e45------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 04:44:07 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@vitd/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[How could the hiring process be revolutionized?]]></title>
            <link>https://medium.com/@vitd/how-could-the-hiring-process-be-revolutionized-e46b5fd51977?source=rss-f48e1b30e45------2</link>
            <guid isPermaLink="false">https://medium.com/p/e46b5fd51977</guid>
            <category><![CDATA[hr]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[recruiting]]></category>
            <category><![CDATA[psychometrics]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Vitalii Duk]]></dc:creator>
            <pubDate>Tue, 28 Apr 2020 17:19:42 GMT</pubDate>
            <atom:updated>2020-04-29T07:44:42.957Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*0oG0QU71R_PD60-z.jpeg" /><figcaption>A proposed solution to the main pain points of the hiring process</figcaption></figure><p>As a seasoned hiring manager, I often reflect on the challenges of the hiring process. Like many of you, I’ve been on both sides — as a candidate as well as a manager — and I see many, many opportunities for improvement. In this article, I discuss how the hiring process could be improved by introducing a better tech platform.</p><p>This platform would be a mixture of an applicant tracking system (ATS) and various assessment platforms.</p><p><a href="#problem-statement"><strong>Problem statement</strong></a></p><p>First, let’s explore the current situation.</p><p><em>How is the recruitment process difficult for the candidate?</em></p><ul><li>The interview process is long and stressful.</li><li>The candidate often doesn’t receive feedback or replies after the pre-screening interview.</li><li>If a candidate applies to several positions in a single company, many steps may be redundant and time-consuming.</li></ul><p><em>And how is the recruitment process challenging for the hiring manager?</em></p><ul><li>Thousands of candidates must be screened, which is a staggering task. And even so, good candidates could be missed.</li><li>It can be hard to eliminate bias and not discriminate based on race, nationality, education, or other characteristics.</li><li>It’s hard to know if a person would be a good fit based just on his CV. On the other hand, you can’t physically interview every potential candidate.</li><li>The many steps of the interview process might be located in various systems: the coding in <a href="https://www.hackerrank.com/">HackerRank</a>; the video interview in <a href="https://www.hirevue.com/">HireVue</a>; the candidate profile in <a href="https://www.greenhouse.io/">Greenhouse</a>; the psychometric test in <a href="https://www.savilleassessment.com/">Saville</a>; and the game-based assessment in <a href="https://www.pymetrics.ai/">pymetrics</a>. It’s hard to track all these components and use the data to make decisions.</li></ul><p><a href="#existing-solutions"><strong>Existing solutions</strong></a></p><p>I could start devising a solution by looking at ATS, such as <a href="https://www.greenhouse.io/">Greenhouse</a>.</p><p><em>Why would I consider these systems?</em></p><ul><li>They have great potential for evolving into multi-purpose platforms.</li><li>Applicant tracking systems feature streamlined integration with various sourcing and job posting platforms, like <a href="https://www.linkedin.com/">LinkedIn</a>, <a href="https://www.indeed.com/">Indeed</a>, and <a href="https://www.glassdoor.com/">Glassdoor</a>.</li><li>Candidates are already interacting with these frontline systems.</li><li>They are already go-to places for hiring managers, despite being not very pleasant from a design standpoint.</li></ul><p><em>Despite their potential, however, most ATS have these weak points:</em></p><ul><li>The initial candidate screening process is barely automated. HR still needs to sift through a huge number of applications, cherry-pick the best, call them, ask trivial questions, and so on.</li><li>UX is overloaded with elements. It’s easy to get lost with hundreds of filters and toggles on a single screen.</li><li>These systems don’t support smooth and native integrations with other systems, such as video interviewing, testing, and screening. This breaks the candidate’s experience, with them getting hundreds of emails with links to various platforms.</li><li>The candidate can’t visualize the number of stages completed and the number of stages left to go.</li><li>The hiring manager has no easy way to see a list of all candidates with scores at different stages of the interview process, clear and simple status indicators.</li><li>The hiring manager also can’t quickly shortlist candidates that pass the hiring criteria.</li><li>Third-party integration of these systems, with the assessment platforms, would be difficult without extensive software engineering skills.</li></ul><p><a href="#proposed-solution"><strong>Proposed solution</strong></a></p><p>As we’re starting with building on top of the ATS, we assume the following things have already been solved:</p><ul><li>Integrations with the sourcing and job posting platforms already exist;</li><li>Interview scheduling capabilities are present;</li><li>There is an ability to send updates to candidates and hiring managers via different mediums, like e-mail and Slack; and</li><li>It has a clean and neat UI and UX, with the ability to easily browse, search, filter, and shortlist candidates.</li></ul><p>Although an ATS would be a great starting point, it wouldn’t give us everything we desire. A few other elements are essential to a smoothly functioning, fully integrated platform.</p><p>Let’s zoom out from the functionality and look at the applicant flow. What would happen after a candidate successfully applied via LinkedIn? In our ideal system, we would welcome our candidate on the main page. There, he could see the positions to which he’s already applied, with the progress and completed number of stages indicated.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*O1tTBx1s4K0j7mNQKUAKEA.png" /><figcaption>Home page</figcaption></figure><p>For each vacancy, there would be a clear indication of which stages the candidate yet needs to complete, indicating the time necessary for completion. This would help the candidate to understand and plan for the entire process.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DvyQCG2PAtY5X-uTBoh-QA.png" /><figcaption>List of interview steps</figcaption></figure><p>Now, consider the stages of the interview flow.</p><ul><li><strong>Screening questions.</strong> These are typical questions that recruiters ask candidates during their initial call. They might include salary expectations, willingness to travel for business, and motivating factors. These questions could easily fit on a single web page to allow a candidate to answer them in one go. If needed, recruiters could set simple filtering rules, excluding candidates that don’t match the criteria; for example, those without a U.S. work visa (H1B).</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nvWVLZEF1TYP6Xu3mTMt5w.png" /><figcaption>Screening questions</figcaption></figure><ul><li><strong>Video interview.</strong> Depending on the vacancy, it might be helpful to assess the communication skills of the candidate via a video interview. Nowadays, many AI-powered platforms claim they can pick up various traits from video images and voice. At the same time, a simple video recording of a person answering a position-specific question could facilitate a decision in the late stages of the interview process. In later posts, I’ll deep-dive into the machine learning algorithms that are available for the video interview analysis.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SIwCU0IRyAdoMJwL8HR9rg.png" /><figcaption>Video interview</figcaption></figure><ul><li><strong>Vacancy-specific tests.</strong> These might be any other professional tests designed specifically for a position or to cater to the specific needs of a company. It could be a typical quiz or something more advanced, such as coding tasks for software engineers (e.g., <a href="https://www.hackerrank.com/">HackerRank</a>). <a href="https://en.wikipedia.org/wiki/Psychometrics">Psychometric</a> and <a href="https://en.wikipedia.org/wiki/Situational_judgement_test">situational judgment tests</a> could fit here as well, as they are multiple-choice questionnaires.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4CYXRLEVTBpCQ-CN9eMsRw.png" /><figcaption>Coding interview</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AjArKcPWsI4RFvuoMjlRmA.png" /><figcaption>Psychometric test</figcaption></figure><ul><li><strong>Game-based assessments</strong>. There are plenty of games that help measure a candidate’s cognitive and emotional attributes and numerical and logical reasoning skills. Start-ups like <a href="https://www.pymetrics.ai/">pymetrics</a> and <a href="https://www.arcticshores.com/">Arctic Shores</a> have been using them successfully. Here are some examples:</li></ul><p>** Those that measure <strong><em>processing speed</em></strong> by asking a candidate to correctly sort falling objects;</p><p>** Those that measure <strong><em>pattern recognition capabilities</em></strong> and problem-solving skills by asking a candidate to find a certain figure among a collection of rotated ones; and</p><p>** Those that assess <strong><em>memory</em></strong> by asking the candidate to reproduce a previously shown path.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*x9_glua0e14QvlU0OE7BaQ.png" /><figcaption>List of the game assessments</figcaption></figure><ul><li><strong>In-person interviews.</strong> These are your typical, face-to-face interviews conducted either in an office or via a video conferencing app. Having these interviews in the same system would allow both the candidate and the hiring manager to keep track of all hiring activities in one system, without switching context.</li></ul><p>The candidate could theoretically go through the entire interview process, all of these steps, without leaving our platform. And, some start-ups have already developed solutions for each component of the flow, and many have been used successfully for several years.</p><p><em>So what’s the problem?</em> There is, as of yet, no platform that combines everything and allows the user to plug-and-play components based on the need for a reasonable price.</p><p>For example, a hiring manager could want a video interview combined with the games, with face-to-face interviews afterward. Of course, some integrators merge several solutions into one flow, but they are expensive, slow, and not scalable.</p><p>Let’s look now at the features the recruiter interface should have. I’ll skip the majority of less relevant admin functionality and focus only on important components.</p><p>The main focus of the recruiter’s UX should be the candidates’ list. This list should have all of the applicants, ranked and shortlisted based on the scoring output, with a multitude of filter and search options.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*E22Sb3JvgKvNrlUAm_hFVA.png" /><figcaption>Candidates list</figcaption></figure><p><a href="#candidate-ranking"><strong>Candidate ranking</strong></a></p><p>How could the output of such diverse assessments be combined, and how could it all flow into the final candidate ranking score?</p><p><em>Let’s break the process down into the following stages:</em></p><ul><li>For each step, let’s develop a conversion function that would take candidate responses as input and spit out a score between 0 and 100, where 100 means a perfect fit. This conversion function could be a simple rule-based engine or a machine learning algorithm, depending on the assessment.</li><li>It would make sense to break down the scoring based on a particular assessment by trait or a specific candidate characteristic, like processing speed. Each assessment in the flow would be responsible for evaluating specific traits.</li><li>For a complex assessment, like games, we would need to run several trials to gather data and build out a baseline scoring system. Those trials would need to include diverse candidate archetypes, not just perfect ones.</li><li>Ranking video assessments could be quite complex, so I’ll address the topic in later articles. For now, assume that our machine learning model for the video interview would output a final score for related traits (e.g., communication abilities) in a range from 0 to 100, where 100 means perfect communication abilities.</li><li>The final score would be calculated based on rules a hiring manager would specify at the beginning while configuring a particular vacancy. A Data Scientist position at a start-up might require candidates with perfect problem solving and stress management skills, so those scores would have the highest weight.</li></ul><p>Our final score would be the key element in our candidate shortlisting experience.</p><p><a href="#avoiding-bias"><strong>Avoiding bias</strong></a></p><p>Hiring processes should be fair and accessible to all qualified applicants.<em> </em>There are several ways to be sure our platform wouldn’t be biased towards a specific gender, nationality, or other factors:</p><ul><li>We could exclude all features from our ML models that could potentially lead to bias, such as gender, age, and nationality.</li><li>By using well-known statistical methods, such as <a href="https://en.wikipedia.org/wiki/Z-test">Z-tests</a> and <a href="https://en.wikipedia.org/wiki/Chi-squared_test">Chi-squared tests</a>, we could constantly check to make sure the predicted fit score is prone to biases. <a href="https://www.pymetrics.ai/">pymetrics</a> released an effective set of Python functions that simplify those checks a lot: <a href="https://github.com/pymetrics/audit-ai">https://github.com/pymetrics/audit-ai</a></li></ul><p><a href="#advantages"><strong>Advantages</strong></a></p><p>So what are the advantages of the proposed system? What problems would it solve?</p><p><em>For hiring managers and companies, it would:</em></p><ul><li>Remove the burden of screening of thousands of candidates, saving time;</li><li>Provide an almost bias-free assessment process;</li><li>Ensure that best candidates are not missed or discarded; and</li><li>Speed up the entire hiring process by an order of magnitude.</li></ul><p><em>For candidates, it would:</em></p><ul><li>Remove gaps between interviews, allowing major part of the flow to be completed in less than two hours;</li><li>Introduce more transparency to the entire process, as candidates could easily see what steps have been done and what is remaining, with time estimates; and</li><li>Give the candidates personalized recommendations of job positions, based on the assessment data.</li></ul><p><a href="#validation"><strong>Validation</strong></a></p><p>A logical question is related to the validation of the approach. Would such a platform be capable of accurately shortlisting candidates?</p><p><em>This could be validated in the following ways:</em></p><ul><li>We could run classic A/B tests, flowing some of the candidates for a particular vacancy through the new platform (treatment group) and others through the existing process (control group). We could collect metrics, like:</li><li>Interview success rates (how many candidates passed through the entire interview process out of those that were shortlisted);</li><li>How many of the new hires passed their probation periods?</li><li>How the new hires performed on their first performance review cycle?</li><li>How many new hires were promoted?</li><li>We could run a test, applying the full assessment of the new platform to existing employees (all or a subset). After obtaining the results, we could compare the predicted fit score to the existing performance data: performance review ratings, promotions, bonuses, and other metrics.</li></ul><p><a href="#final-words"><strong>Final words</strong></a></p><p>The platform I envision would resolve a variety of hiring process issues by providing a streamlined, user-friendly, powerful hub. Through the entire cycle, both the client manager and the candidate would be interacting with one piece of software, one that is 100% transparent and clear. That would make the hiring process much more pleasant and efficient.</p><p>In general, it would be great to make such a platform open source, so organizations could easily set up and install it independently, without needing to spend a lot of money. I see it working much like <a href="https://github.com/wikimedia/mediawiki">MediaWiki</a>.</p><p>For the past several months, I’ve been working on the separate components of such a system, which I plan to land on my <a href="https://github.com/root-ua">Github</a> account, soon, for free.</p><p><em>I welcome your feedback and ideas on this topic.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e46b5fd51977" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to fail at building a Data Science team]]></title>
            <link>https://medium.com/@vitd/how-to-fail-at-building-a-data-science-team-8e7d7c20a51d?source=rss-f48e1b30e45------2</link>
            <guid isPermaLink="false">https://medium.com/p/8e7d7c20a51d</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[experience]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Vitalii Duk]]></dc:creator>
            <pubDate>Wed, 08 Apr 2020 14:28:48 GMT</pubDate>
            <atom:updated>2020-04-15T20:07:48.871Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/500/0*zw1qYNN9K9b-Ur2v.png" /><figcaption>9 easy steps to prevent your organization from doing actual Data Science</figcaption></figure><p><strong>Disclaimer:</strong> <em>The following post is loaded with sarcasm, so take it with a grain of salt and be mindful :)</em></p><p>Over the past several years in the industry, I’ve seen many companies trying to build their Data Science, or how they call it, AI departments, from scratch. Most of them were quite ambitious. Most of them failed.</p><p>Based on some observations, I’ve put together a list of “best practices” that, when applied at your company, can quite certainly result in failed attempts to build a Data Science competence. Please follow those.</p><p>So, where do you begin?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/406/0*ExCg23qE3jwAx2x4.gif" /></figure><p><strong>Be ambitious</strong></p><p>Start with grand ambitions. Plan to hire at least 100 Data Scientists and Machine Learning Engineers by the end of the year. It’s essential to market such initiatives with pumped up news headlines on LinkedIn.</p><p>It doesn’t matter that currently, you have zero Data Scientists in total. As a matter of fact, it doesn’t matter whether you have a data warehouse, proper data ingestion infrastructure, or slightest understanding of what business problems you’re solving. AI <a href="https://www.urbandictionary.com/define.php?term=FTW">ftw</a>! Be ambitious.</p><p>Such an attitude may be an excellent way to scare off your competition.</p><p><strong>Get consultants onboard</strong></p><p>Hiring someone from the <a href="https://en.wikipedia.org/wiki/Big_Three_(management_consultancies)">Big Three</a> is always a good idea. Yes, it’s expensive, but on the other hand, they take on all responsibility for any failures of your Data and AI strategy. These guys know what to do, as they’ve delivered millions of similar presentations across the globe.</p><p>Oh wait, did I say presentations and not actual projects? Well, it doesn’t matter. At the end of the day, what’s more important? Strategy, right? You’ll quickly figure out how to execute and implement each point from that 110-page deck later on. On your own.</p><p>Wait, is there some smart-ass on your team who disagrees with high-profiled consultants? They asked to explain how a number on slide 34 got derived, and the experts couldn’t come up with an answer? You should learn some manners!</p><p><strong>Solve every business problem with AI</strong></p><p>So you’ve heard about the tremendous success of AI in various industries, like <a href="https://deepmind.com/research/case-studies/alphago-the-story-so-far">DeepMind’s AlphaGo</a>, that can beat a human at classic Go game or <a href="https://en.wikipedia.org/wiki/Libratus">Libratus</a> that is extremely good in Texas Hold ’Em, right?</p><p>Then why don’t you apply some of those algorithms to solve your business problem? Or maybe all of them? It doesn’t matter if you have good enough quality data or not. AI can solve everything.</p><p><strong>Use only the cutting-edge tech</strong></p><p>Let’s say you’ve just read about some cool Machine Learning algorithms developed by the <a href="https://ai.facebook.com/">Facebook FAIR team</a>. Have you learned anything about <a href="https://en.wikipedia.org/wiki/Long_short-term_memory">LSTMs</a>? Go ahead and tell your team to use it immediately!</p><p>First, your team will appreciate your guidance in this area and the urge to apply the best existing solutions out there. Second, most likely, the use of cutting-edge algorithms will solve all your problems. In the end, isn’t AI smart enough to sort out those data quality issues you have?</p><p><strong>Ph.D. is the key</strong></p><p>Hire only Ph.D.s, preferably those who are deep into academic research. Getting a Ph.D. can take a long time, longer than in many other fields, which automatically equals persistence and dedication in getting things done.</p><p>Ph.D. guys will enjoy building pre-processing data pipelines, cleaning your messy and incomplete data, and creating dashboards for you in Tableau. In other words — applying AI. Isn’t that what they did while getting their Ph.D.?</p><p><strong>Call everything AI</strong></p><p>Have you built a simple rule-based engine to detect some basic fraud happening on your platform? Call it, AI! Do you have a simple linear regression to predict prices? It’s definitely AI! Did you do a simple data wrangling in Python? Undoubtedly, you’re applying AI to your decision making.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/0*vWIWHvJo9PdUob4G.png" /></figure><p>As long as you feel knowledgeable and empowered by calling your processes according to the latest tech solutions to cover your gaps, it doesn’t matter what stands behind it. Otherwise, why is everyone uses AI if it doesn’t work?</p><p><strong>Take it slow</strong></p><p>Do your projects get stuck for months in the R&amp;D phase? Have you just spent the last few months fine-tuning the hyper-parameters and improving your precision score by 0.1%? No worries, it’s all good, there is no need to rush.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/320/0*vmj0cB3LkzdoEdAC.gif" /></figure><p>Or maybe you’re wondering what might happen if prerequisites of the project, data distribution, initial assumptions, and business requirements change by the time you finish all the phlegmatic improvements? You start over again. It’s as simple as that.</p><p><strong>Bureaucracy runs the world</strong></p><p>If you want to improve your Data Science team productivity, there are a few things you can do:</p><ul><li><em>Does someone want to install additional Python package?</em> Let them create a ticket for IT support and wait for two weeks for the approval.</li><li><em>Did you get a request for work from home or more flexible working hours?</em> Everyone should be in the office by 7:30 AM every day! Okay, fine, by 8 AM on Thursdays.</li><li><em>Do your Data Scientists wear t-shirts and jeans to work?</em> A suite is a must! Okay, maybe not as expensive as the one on your consultants, but still.</li></ul><p><strong>Invest in drag-n-drop</strong></p><p>Have you seen those fancy AI tools with friendly UX/UI that allow you to drag-n-drop Machine Learning models straight into production? Buy all of them! Is someone telling you don’t need them? Let’s figure out the details later.</p><p>Yes, it’s quite expensive, but aren’t they cute and lovely? Your C-level stakeholders will, for sure, appreciate such an upgrade. Otherwise, how are you going to present your results? Using those ugly <a href="https://jupyter.org/">Jupyter Notebooks</a>? No way!</p><p><em>Do you know any other best practices that will help fail the company’s Data Science initiatives? Feel free to share your insights and “expertise”.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8e7d7c20a51d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[9 Cataclysmic Ways to Make Your Tech Startup Fail]]></title>
            <link>https://medium.com/@vitd/9-cataclysmic-ways-to-make-your-tech-startup-fail-1de012ae3d5c?source=rss-f48e1b30e45------2</link>
            <guid isPermaLink="false">https://medium.com/p/1de012ae3d5c</guid>
            <category><![CDATA[knowledge-sharing]]></category>
            <category><![CDATA[fashion-technology]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Vitalii Duk]]></dc:creator>
            <pubDate>Thu, 01 Aug 2019 04:01:01 GMT</pubDate>
            <atom:updated>2019-08-01T07:13:10.451Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VKZFlAwBZCl66MMmzStUAg.jpeg" /><figcaption>How I failed in my tech startup — and how you can do it too.</figcaption></figure><p>Building a startup is one of the easiest things you can do.</p><p>But failing your startup?</p><p>That takes willful ignorance and a complete lack of insightful direction.</p><p>And frankly, just not everyone’s cut out for that kinda hustle.</p><p>I was lucky enough to have the non-existent expertise to push through my better judgment and completely and utterly fail my fashion tech startup last year.</p><p>Now I’m ready to be internet-vulnerable and pass on some very legitimate advice…</p><p>And show you how you, too, can be a complete and utter startup failure.</p><p>Okay.</p><p>Here goes something.</p><p><strong>1. Have Only Technical Co-Founders</strong></p><p>Yes, you could do the smart thing and have a balanced mix of technical and non-technical leadership…</p><p>But c’ mon, where’s the fun in that?</p><p>So, the key is to make sure all your founders are tech guys.</p><p>They can easily learn all the business stuff they need to know on YouTube or whatever and besides, even if you had someone with the business skills, how would you communicate in anything other than Python? Is it even fun talking to someone who can’t take a derivative?</p><p>People will tell you to change your course on this, but you keep going.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/380/1*9ds501eC4qta8Wmbl36T-w.gif" /><figcaption><em>Ignore the haters.</em></figcaption></figure><p><strong>2. Rush the Product Build</strong></p><p>There are three keys to your product after you formulate a basic idea:</p><p>· Don’t waste time thinking about your business plan, marketing strategy, or future profitability</p><p>· Build code, code all night long, what are we coding for, I dunno, but keep going</p><p>· Speak with your potential (read: potentially non-existent) customers later on</p><p><em>Yeah, this is just how the Facebook guys did it!</em></p><p><strong>3. Don’t Worry About Your Legal Setup</strong></p><p>Seriously, chill. Legal schmegal. You need to stop worrying about it.</p><p>A shareholders’ or IP assignment agreement with friends or former colleagues?</p><p><em>Easy peasy.</em></p><p>Use the lawyer’s money to chat about these definitely-not-acrimonious details over a pint of beer.</p><p>At the end of the day, what could possibly go wrong?</p><p><strong>4. Equity</strong></p><p>You may be tempted to give up some equity for advisors, developers, digital marketing experts, or, well, anyone who may be looking to join your startup in the early stages to help you accelerate towards success but…</p><p><strong>IGNORE THEM.</strong></p><p>They are greedy. They will only help you grow fast.</p><p>And you sure as heck don’t need any of that. No, sir.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*-IxMlbhBP8tT7_uWF185BQ.gif" /><figcaption><em>Let the greed fuel you, let it fail you.</em></figcaption></figure><p>While 2% of zero equity is still zero, it’s <strong>your</strong> zero. You made that failed startup baby completely on your own and will reap the rock-bottom rewards of a year-long depression post-fail because of it.</p><p>You’re the founders, you have the idea — and you know better than anyone already in the market.</p><p>So keep that equity all to yourselves.</p><p><strong>5. Details</strong></p><p>Worry over every. Last. One.</p><p>Every button? Perfectly aligned.</p><p>Search bar? That is too long. Fix it. Miss that deadline.</p><p>Even if this slows you down… sets you off course… puts you into bankruptcy…</p><p><em>Keep going.</em> Be the perfectionist you want to see in the world.</p><p><strong>6. Metrics</strong></p><p>Only measure something that looks impressive, like app installs, rather than DAUs and retention.</p><p>If the sign-ups leave or aren’t engaged…</p><p>All good. :)</p><p>Investors won’t be asking questions, anyway — they’re only looking to burn money.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*vIsM0XEGMI8GgWy1rl2zmA.gif" /><figcaption><em>We don’t need good metrics where we’re going.</em></figcaption></figure><p><strong>7. Take Your Time</strong></p><p>Building and iterating fast to be first to market is for schmucks.</p><p>Startup schmucks who succeed.</p><p>You definitely don’t want to be one of those, so be sure to take your time and not worry about if someone else is building and releasing a similar product to yours.</p><p><strong>8. Stick to Your Core Skills</strong></p><p>Software engineers were born to code — not to worry about kinda-crucial things like digital marketing and startup strategy.</p><p>But in a lean startup environment that requires all hands on business development, coding, and pitching, it’s best to not learn anything else outside of your scope.</p><p>Instead, keep your head down stubbornly and watch it all burn.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/404/1*SCFecOBp8mV7Slee6JAwXw.gif" /></figure><p><strong>9. Tunnel Vision</strong></p><p>Ah, yes. Operating with blinders on — my favorite tip.</p><p>If you created a startup around your own problem — not the market’s — and see warning signs, or don’t even bother looking for any until you’ve built a product, you’re doing it all right.</p><p>And I just want to say how proud I am of you for failing.</p><p>***</p><p>Know any other best practices to help fail a startup? Share your “expertise” below.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1de012ae3d5c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>