<?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: nghiep pham quoc</title>
    <description>The latest articles on DEV Community by nghiep pham quoc (@nghiep_phamquoc_568df0fe).</description>
    <link>https://dev.to/nghiep_phamquoc_568df0fe</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%2F4242856%2F84512b22-c3d9-4e86-8008-4cbdbda6e303.png</url>
      <title>DEV Community: nghiep pham quoc</title>
      <link>https://dev.to/nghiep_phamquoc_568df0fe</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/nghiep_phamquoc_568df0fe"/>
    <language>en</language>
    <item>
      <title>Why I'm building Lab to Prod in public</title>
      <dc:creator>nghiep pham quoc</dc:creator>
      <pubDate>Sat, 14 Nov 2026 07:34:40 +0000</pubDate>
      <link>https://dev.to/nghiep_phamquoc_568df0fe/why-im-building-lab-to-prod-in-public-1pcl</link>
      <guid>https://dev.to/nghiep_phamquoc_568df0fe/why-im-building-lab-to-prod-in-public-1pcl</guid>
      <description>&lt;p&gt;20 posts, 8 short videos and 4 long articles in 4 weeks, with at most 30 minutes of my own time&lt;br&gt;
per week. That is the target of the experiment that starts on Monday 16 November 2026. This&lt;br&gt;
article explains why I am running it, why the whole thing is public, and what would make me stop.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where the name comes from
&lt;/h2&gt;

&lt;p&gt;I trained as a chemist. In a lab, a result does not count because you saw it once. It counts&lt;br&gt;
when someone else can follow your notebook and get the same number. Most of the work is not the&lt;br&gt;
experiment. It is the controls, the calibration and the notes.&lt;/p&gt;

&lt;p&gt;Today I build AI systems, and the distance between a demo and a system you can trust feels very&lt;br&gt;
familiar. A prompt that works once on your laptop is a result you saw once. Production is the&lt;br&gt;
part where you add the controls: evaluation sets, quality gates, logs, a way to roll back.&lt;/p&gt;

&lt;p&gt;"Lab to Prod" is that path. It is also my own path, from a lab bench to production software, and&lt;br&gt;
I am still on it. I am studying data structures in Java, linear algebra, cloud architecture and&lt;br&gt;
climate science at the same time, and I forget things I do not write down.&lt;/p&gt;

&lt;h2&gt;
  
  
  Three reasons, in order
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;1. To show the work.&lt;/strong&gt; Saying "I know LLM evaluation" is cheap. A dated trail of notes,&lt;br&gt;
code and measurements is harder to fake and easier to check. I would rather be judged on that.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. To learn properly.&lt;/strong&gt; Explaining something in 200 words to another engineer is the fastest&lt;br&gt;
way I know to find out that I do not understand it yet. Each note forces one idea to be clear&lt;br&gt;
enough to survive a stranger reading it.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;3. Because the pipeline is the project.&lt;/strong&gt; I work on AI engineering: agents, evaluation,&lt;br&gt;
human approval gates, monitoring. A publishing pipeline needs all four. So instead of describing&lt;br&gt;
those ideas in the abstract, I am building a small real system that uses them, and you can read&lt;br&gt;
every line of it.&lt;/p&gt;

&lt;h2&gt;
  
  
  What "in public" means here
&lt;/h2&gt;

&lt;p&gt;Everything lives in &lt;a href="https://github.com/nghiep-pham-ecosys/lab-to-prod" rel="noopener noreferrer"&gt;one repository&lt;/a&gt;: the calendar,&lt;br&gt;
every post as a markdown file, the code that draws the images and videos, the scripts that&lt;br&gt;
publish, and the metrics that come back.&lt;/p&gt;

&lt;p&gt;The weekly loop is short:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;On Saturday, a Claude Code session drafts next week's five notes, their images, two short
videos and one article, and opens a single pull request.&lt;/li&gt;
&lt;li&gt;A validation job checks that pull request: length limits per channel, image and video sizes,
dates against the calendar, source links for the topics where numbers are easy to get wrong,
and a list of terms that must never appear.&lt;/li&gt;
&lt;li&gt;I read it on my phone. I edit, comment, or merge.&lt;/li&gt;
&lt;li&gt;After the merge, GitHub Actions publishes each post on its day through each platform's
official API and records the result, so a re-run never posts twice.&lt;/li&gt;
&lt;li&gt;Every night, a job collects the numbers. Every Sunday, a report compares them with the
targets and proposes one change for the following week.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;The rule that matters most is in step 3: &lt;strong&gt;no merge, no posts.&lt;/strong&gt; If I am busy and the pull&lt;br&gt;
request sits there, the week is skipped. That is deliberate. An automated pipeline that&lt;br&gt;
publishes without a human reading it is a way to be wrong at scale.&lt;/p&gt;

&lt;p&gt;As I write this, the repository holds the calendar, the validator, the brand assets and this&lt;br&gt;
blog. The publishers are not written yet. You are reading the first entry of the notebook, not&lt;br&gt;
a victory lap.&lt;/p&gt;

&lt;h2&gt;
  
  
  The constraints
&lt;/h2&gt;

&lt;p&gt;A few limits shape the design, and I think they make it more interesting:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Zero budget.&lt;/strong&gt; No paid APIs anywhere. Where a platform charges for posting through its API,
the pipeline writes a file and I schedule that post by hand. It costs me about five minutes a
week.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Nothing from work.&lt;/strong&gt; No code, data, numbers or stories from any employer or client. Every
example is either measured in this repository, with the script next to it, or labelled as a
toy example.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;I answer people myself.&lt;/strong&gt; Replies and messages are never automated. If you write to me, you
get me.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A kill switch.&lt;/strong&gt; One file named &lt;code&gt;PAUSE&lt;/code&gt; at the root of the repository stops every publishing
job. I can create it from my phone.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Corrections stay visible.&lt;/strong&gt; If a post is wrong, it gets a correction note. It does not get
deleted.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  How I will know if it works
&lt;/h2&gt;

&lt;p&gt;These are the numbers I committed to before starting. I am writing them down now so I cannot&lt;br&gt;
move them later.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Measure&lt;/th&gt;
&lt;th&gt;Target after 4 weeks&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Published on schedule&lt;/td&gt;
&lt;td&gt;at least 90% (18 of 20 notes, 7 of 8 videos)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;My time&lt;/td&gt;
&lt;td&gt;at most 30 minutes a week&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Posts corrected after publishing for a technical error&lt;/td&gt;
&lt;td&gt;0&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Average LinkedIn impressions per post&lt;/td&gt;
&lt;td&gt;at least 300&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;New followers across all channels&lt;/td&gt;
&lt;td&gt;100&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Real conversations&lt;/td&gt;
&lt;td&gt;at least 10 comments and 1 direct message from an engineer&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The engagement targets are guesses. I will adjust them after the first week of real data, and I&lt;br&gt;
will say so when I do. Five posts a week is also a very small sample, so the weekly experiments&lt;br&gt;
(a different hook, a different format, a different posting time) will produce hints, not&lt;br&gt;
conclusions.&lt;/p&gt;

&lt;p&gt;At the end I will score the trial on six criteria: regularity, my effort, technical accuracy,&lt;br&gt;
growth, quality of conversations, and whether I actually learned more. Above 3.5 out of 5, it&lt;br&gt;
continues. Between 2.5 and 3.4, it slows down to three posts a week. Below 2.5, it stops, and I&lt;br&gt;
keep whatever parts were useful.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I do not know yet
&lt;/h2&gt;

&lt;p&gt;Three things worry me, and I would rather name them now.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Can a model grade its own drafts?&lt;/strong&gt; From week three, I want to try silent approval: the weekly&lt;br&gt;
pull request merges itself if validation passes and a separate grading step, run by a model that&lt;br&gt;
did not see the drafting, scores every post at least 4 out of 5. I do not know yet whether that&lt;br&gt;
grade means anything. That question gets its own article.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Will the writing sound like me?&lt;/strong&gt; Drafts come from a model. I edit them, and I am responsible&lt;br&gt;
for every sentence, but voice drifts. I will keep comparing against posts I wrote entirely by&lt;br&gt;
hand.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Is daily too much?&lt;/strong&gt; One post per channel per day is the plan. If the numbers say people tune&lt;br&gt;
out, the cadence changes.&lt;/p&gt;

&lt;h2&gt;
  
  
  What to expect
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Monday to Friday:&lt;/strong&gt; one short note. Monday is data structures in Java, Tuesday is ML
foundations, Wednesday is AI engineering, Thursday is cloud and DevOps, Friday alternates
between climate and simulation.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Saturday:&lt;/strong&gt; one long article. Either a build log like this one, with the week's numbers and
failures, or a deeper version of the best note of the week.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you want to follow along, the &lt;a href="https://nghiep-pham-ecosys.github.io/lab-to-prod/rss.xml" rel="noopener noreferrer"&gt;RSS feed&lt;/a&gt; has everything. If you spot a&lt;br&gt;
mistake, open an issue on the repository. That is exactly the kind of review a lab notebook is&lt;br&gt;
for.&lt;/p&gt;




&lt;p&gt;&lt;em&gt;Originally published on &lt;a href="https://nghiep-pham-ecosys.github.io/lab-to-prod/posts/why-i-am-building-lab-to-prod-in-public/" rel="noopener noreferrer"&gt;Lab to Prod&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>buildinpublic</category>
      <category>softwaredevelopment</category>
    </item>
  </channel>
</rss>
