<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Ines Mota</title>
    <description>The latest articles on DEV Community by Ines Mota (@ines_mota).</description>
    <link>https://dev.to/ines_mota</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4215044%2F76004b6f-e70c-4642-bdfa-616c14c83f39.jpg</url>
      <title>DEV Community: Ines Mota</title>
      <link>https://dev.to/ines_mota</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/ines_mota"/>
    <language>en</language>
    <item>
      <title>A checklist before you ship a notebook model</title>
      <dc:creator>Ines Mota</dc:creator>
      <pubDate>Sun, 01 Nov 2026 15:07:36 +0000</pubDate>
      <link>https://dev.to/ines_mota/a-checklist-before-you-ship-a-notebook-model-2bn4</link>
      <guid>https://dev.to/ines_mota/a-checklist-before-you-ship-a-notebook-model-2bn4</guid>
      <description>&lt;p&gt;I've shipped enough models from notebooks to know that the notebook is the least of it. Six checks I run before anything goes to a service:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Snapshot the data.&lt;/strong&gt; The exact table or file, with a hash. "It was in BigQuery" is not reproducible.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Freeze an eval set.&lt;/strong&gt; Not the test split from training — a small, ugly, real-world one.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Check the input distribution.&lt;/strong&gt; Compare the serving inputs to the training inputs; the drift is usually in the tail.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Measure the serving path, not the model.&lt;/strong&gt; p50 and p99 end to end.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Write down the failure behavior.&lt;/strong&gt; What returns when the model is uncertain, slow, or wrong?&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Plan the rollback.&lt;/strong&gt; A model that can't be turned off in minutes is a liability.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;None of this is novel. It's just the list I forget when I'm excited about the notebook.&lt;/p&gt;

&lt;p&gt;What's on your pre-ship list?&lt;/p&gt;

</description>
      <category>machinelearning</category>
      <category>python</category>
      <category>mlops</category>
    </item>
    <item>
      <title>The boring half of shipping a model</title>
      <dc:creator>Ines Mota</dc:creator>
      <pubDate>Thu, 29 Oct 2026 10:47:42 +0000</pubDate>
      <link>https://dev.to/ines_mota/the-boring-half-of-shipping-a-model-doj</link>
      <guid>https://dev.to/ines_mota/the-boring-half-of-shipping-a-model-doj</guid>
      <description>&lt;p&gt;The model was the easy part. It took two weeks to train something good enough, and about five more to make it something the team could rely on.&lt;/p&gt;

&lt;p&gt;The half nobody puts in the writeup:&lt;/p&gt;

&lt;p&gt;Data validation before inference. If a field arrives empty or in a new format, the model still returns a confident answer. We added a schema check and a rule: if validation fails, return a clear error instead of a guess.&lt;/p&gt;

&lt;p&gt;A version pair. Every prediction logs the model version and the preprocessing version, because they drift independently and a mismatch produces results that look plausible.&lt;/p&gt;

&lt;p&gt;An eval set that lives in the repo and runs in CI. Three hundred labeled examples, run on every change. It has caught two 'harmless' preprocessing tweaks that quietly cost four points.&lt;/p&gt;

&lt;p&gt;A cost number. Not accuracy, not F1, just the price per thousand documents. It shapes the design more than any other metric.&lt;/p&gt;

&lt;p&gt;None of this is interesting at a conference. It is the difference between a demo and a service.&lt;/p&gt;

</description>
      <category>machinelearning</category>
      <category>mlops</category>
      <category>python</category>
    </item>
  </channel>
</rss>
