<?xml version="1.0"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title><![CDATA[Sagar Kapoor]]></title><link>https://sagarkapoor.eu/</link><description><![CDATA[Sagar Kapoor. I live in Sofia and build software for the web. I’m also into trading, Kabbalah, and making games.]]></description><language>en</language><lastBuildDate>Thu, 19 Nov 2026 17:31:19 GMT</lastBuildDate><docs>https://www.rssboard.org/rss-specification</docs><item><title><![CDATA[I'd rather quote a timeline I believe]]></title><link>https://sagarkapoor.eu/writing/honest-timelines</link><description><![CDATA[Padding an estimate to win the deal doesn't help anyone for long. Here's how I think about quoting dates — and why I'd rather lose a pitch than miss one.]]></description><category>studio</category><guid isPermaLink="true">https://sagarkapoor.eu/writing/honest-timelines</guid><pubDate>Mon, 18 May 2026 00:00:00 GMT</pubDate><content:encoded><![CDATA[<p>There&#39;s a familiar pressure when a client asks &quot;how long will this take?&quot; The honest answer might be twelve weeks, but six is the number that wins the work. It&#39;s tempting — I&#39;ve felt the pull.</p>
<p>I try not to give in to it, and not because I&#39;m above it. I&#39;ve just watched what the short number does: you win the deal, then spend the next three months managing disappointment instead of building. Nobody enjoys that stretch, least of all the client.</p>
<h2 id="the-date-i-actually-believe">The date I actually believe <a class="heading-anchor" href="https://sagarkapoor.eu/writing/honest-timelines#the-date-i-actually-believe" aria-label="Link to section: The date I actually believe">#</a></h2>
<p>When I quote a timeline, I aim for the date I genuinely think we&#39;ll hit, with some room for the things that usually go sideways. Sometimes that makes my number bigger than the next quote, and sometimes I lose the work because of it. I&#39;ve made peace with that trade.</p>
<p>A few habits keep me honest:</p>
<ul>
<li>Saying no to work I can&#39;t do well in the time we have.</li>
<li>Flagging early when something turns out harder than it looked.</li>
<li>Trying not to bill for my own uncertainty.</li>
</ul>
<blockquote>
<p>An estimate is a quiet promise. If I wouldn&#39;t make the promise out loud, I try not to hide it inside a number.</p>
</blockquote>
<h2 id="buffer-don-t-inflate">Buffer, don&#39;t inflate <a class="heading-anchor" href="https://sagarkapoor.eu/writing/honest-timelines#buffer-don-t-inflate" aria-label="Link to section: Buffer, don't inflate">#</a></h2>
<p>There&#39;s a real difference between leaving room for known unknowns and padding to cover myself. The first is just planning. The second is charging someone for my nerves, and I&#39;d rather not.</p>
<p>So I&#39;m occasionally not the cheapest quote in the room. That&#39;s a trade I&#39;m comfortable with — I&#39;d rather under-promise and show up than win on a date I can&#39;t keep.</p>
]]></content:encoded></item><item><title><![CDATA[A boring stack is usually a fast one]]></title><link>https://sagarkapoor.eu/writing/boring-stack-fast-sites</link><description><![CDATA[Nuxt, markdown, a CDN. The least exciting choices tend to be the ones that quietly hold up — here's why I keep reaching for them.]]></description><category>engineering</category><guid isPermaLink="true">https://sagarkapoor.eu/writing/boring-stack-fast-sites</guid><pubDate>Thu, 02 Apr 2026 00:00:00 GMT</pubDate><content:encoded><![CDATA[<p>Some of the fastest sites I&#39;ve built were also the most boring to describe. Nuxt. Markdown. A CDN. A server route or two where I actually needed one. It isn&#39;t a clever setup, and that&#39;s sort of the point.</p>
<h2 id="clever-has-a-cost">Clever has a cost <a class="heading-anchor" href="https://sagarkapoor.eu/writing/boring-stack-fast-sites#clever-has-a-cost" aria-label="Link to section: Clever has a cost">#</a></h2>
<p>Every interesting dependency is a small promise to my future self: <em>I&#39;ll keep understanding this.</em> I don&#39;t always keep it, and then the clever thing becomes the part of the codebase nobody wants to touch. Boring tools are just easier to live with — easy to read, easy to replace.</p>
<p>That&#39;s not a knock on anyone reaching for newer, shinier things; plenty of them are great, and I reach for them too when they earn their place. It&#39;s only where I&#39;ve landed after enough late nights debugging my own cleverness.</p>
<pre class="shiki shiki-themes github-light-high-contrast github-dark-high-contrast" style="--shiki-light:#0e1116;--shiki-dark:#f0f3f6;--shiki-light-bg:#ffffff;--shiki-dark-bg:#0a0c10" tabindex="0"><code><span class="line"><span style="--shiki-light:#66707B;--shiki-dark:#BDC4CC">// The most exciting line in the whole codebase is a prerender rule.</span></span>
<span class="line"><span style="--shiki-light:#702C00;--shiki-dark:#FFB757">routeRules</span><span style="--shiki-light:#0E1116;--shiki-dark:#F0F3F6">: {</span></span>
<span class="line"><span style="--shiki-light:#032563;--shiki-dark:#ADDCFF">  "/**"</span><span style="--shiki-light:#0E1116;--shiki-dark:#F0F3F6">: { prerender: </span><span style="--shiki-light:#023B95;--shiki-dark:#91CBFF">true</span><span style="--shiki-light:#0E1116;--shiki-dark:#F0F3F6"> },</span></span>
<span class="line"><span style="--shiki-light:#0E1116;--shiki-dark:#F0F3F6">}</span></span></code></pre><h2 id="ship-it-to-the-edge-and-relax">Ship it to the edge and relax <a class="heading-anchor" href="https://sagarkapoor.eu/writing/boring-stack-fast-sites#ship-it-to-the-edge-and-relax" aria-label="Link to section: Ship it to the edge and relax">#</a></h2>
<p>When the HTML is already sitting on a CDN near the reader, you don&#39;t need much to make it feel fast — it just is. The interesting work is deciding what genuinely has to run at request time (a contact form, an auth check) and letting everything else be static.</p>
<ul>
<li>Prerender what you can.</li>
<li>Run on the edge what you must.</li>
<li>Reach for a database when a file honestly won&#39;t do.</li>
</ul>
<p>The least flashy option on each of those lines is, more often than not, the one I end up glad I picked.</p>
]]></content:encoded></item><item><title><![CDATA[Ship first, then talk about it]]></title><link>https://sagarkapoor.eu/writing/ship-then-talk</link><description><![CDATA[Open code is the most honest pitch I know. Read it, fork it, judge it — that's the kind of trust I'd want before paying anyone.]]></description><category>open-source</category><guid isPermaLink="true">https://sagarkapoor.eu/writing/ship-then-talk</guid><pubDate>Sat, 21 Feb 2026 00:00:00 GMT</pubDate><content:encoded><![CDATA[<p>One of the more honest things a developer can do is just hand you the code. Not a case study, not a testimonial — the actual files, on GitHub, with the rough edges still showing.</p>
<h2 id="trust-you-can-clone">Trust you can clone <a class="heading-anchor" href="https://sagarkapoor.eu/writing/ship-then-talk#trust-you-can-clone" aria-label="Link to section: Trust you can clone">#</a></h2>
<p>Anyone can write &quot;we care about quality&quot; on a homepage; I&#39;ve written something like it myself. It tends to mean more when you can go read the commits. So we try to build our tools in the open — the particle field behind a hero, the carousel on a client site — partly because it keeps us honest, and partly because it lets you check the care is real before you commit to anything.</p>
<p>That&#39;s the relationship I hope for with clients: look at the work first, decide later. And if the open code doesn&#39;t win you over, that&#39;s completely fair.</p>
<h2 id="talking-comes-after">Talking comes after <a class="heading-anchor" href="https://sagarkapoor.eu/writing/ship-then-talk#talking-comes-after" aria-label="Link to section: Talking comes after">#</a></h2>
<p>It&#39;s tempting to announce things before they really exist. I&#39;ve slowly learned to flip the order:</p>
<ol>
<li>Build the thing.</li>
<li>Use it on something real.</li>
<li><em>Then</em> write about it.</li>
</ol>
<p>By the time I&#39;m saying a tool is good, it&#39;s usually already holding something up. The writing is a postscript, not a pitch.</p>
]]></content:encoded></item></channel></rss>