<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Sciences44 Blog</title>
    <description>Insights and tutorials on data engineering, machine learning, and scientific computing.</description>
    <link>https://sciences44.com</link>
    <atom:link href="https://sciences44.com/blog/rss" rel="self" type="application/rss+xml"/>
    <language>en-us</language>
    <lastBuildDate>Sat, 29 Aug 2026 23:52:21 GMT</lastBuildDate>
    <generator>Sciences44 Next.js</generator>
    <image>
      <url>https://sciences44.com/logo.png</url>
      <title>Sciences44 Blog</title>
      <link>https://sciences44.com</link>
    </image>
    
    <item>
      <title><![CDATA[Distill Your Docs: an 8B Teacher, a 0.6B Student, and What Actually Transfers]]></title>
      <link>https://sciences44.com/blog/distill-your-docs-8b-teacher-06b-student-on-your-mac</link>
      <guid isPermaLink="true">https://sciences44.com/blog/distill-your-docs-8b-teacher-06b-student-on-your-mac</guid>
      <description><![CDATA[An 8B reasoning model summarizes and labels documents for a retrieval index very well, at 10 seconds per document. I distilled the task into two 0.6B students on a Mac, each taught by a different 8B teacher. One became a perfect classifier, the other a faithful writer. Neither got both, and that is the real lesson.]]></description>
      <pubDate>Sun, 12 Jul 2026 00:00:00 GMT</pubDate>
      <author>Adham Sersour</author>
      <category>ai</category>
      <category>generative-ai</category>
      <enclosure url="/media/blog_images/distill-your-docs-cover.jpg" type="image/jpeg"/>
    </item>
    <item>
      <title><![CDATA[Is Your Agent's Memory Worth Its Tokens?]]></title>
      <link>https://sciences44.com/blog/is-your-agents-memory-worth-its-tokens</link>
      <guid isPermaLink="true">https://sciences44.com/blog/is-your-agents-memory-worth-its-tokens</guid>
      <description><![CDATA[Agent memory is often measured against a weak baseline. Give a plain agent the same token budget and the advantage mostly disappears into seed variance. Here is a small, reproducible experiment on a local model, and the measurement discipline it teaches.]]></description>
      <pubDate>Tue, 07 Jul 2026 00:00:00 GMT</pubDate>
      <author>Adham Sersour</author>
      <category>ai</category>
      <category>generative-ai</category>
      <enclosure url="/media/blog_images/agent-memory-worth-its-tokens-cover.jpg" type="image/jpeg"/>
    </item>
    <item>
      <title><![CDATA[Scaffold Engineering: Why the Code Around Your LLM Decides More Than the Model]]></title>
      <link>https://sciences44.com/blog/scaffold-engineering-why-the-code-around-your-llm-decides-more-than-the-model</link>
      <guid isPermaLink="true">https://sciences44.com/blog/scaffold-engineering-why-the-code-around-your-llm-decides-more-than-the-model</guid>
      <description><![CDATA[Frontier coding LLMs have converged within 0.8 points on SWE-bench Pro. The scaffold around them is now the dominant performance variable. Here is what scaffold engineering means, the data behind the claim, and what I found running it locally on Qwen3.6-35B-A3B.]]></description>
      <pubDate>Mon, 01 Jun 2026 00:00:00 GMT</pubDate>
      <author>Adham Sersour</author>
      <category>ai</category>
      <category>generative-ai</category>
      <enclosure url="/media/blog_images/scaffold-engineering-cover-v2.png" type="image/jpeg"/>
    </item>
    <item>
      <title><![CDATA[A Pedagogical Guide to GenAI Observability, GenAI Evaluation, and GenAI Monitoring]]></title>
      <link>https://sciences44.com/blog/a-pedagogical-guide-to-genai-observability-genai-evaluation-and-genai-monitoring</link>
      <guid isPermaLink="true">https://sciences44.com/blog/a-pedagogical-guide-to-genai-observability-genai-evaluation-and-genai-monitoring</guid>
      <description><![CDATA[Unlock trustworthy AI: Master evaluation, monitoring, and observability. Discover why DevOps thinking fails AI and how a "helix" model elevates your systems. Stop firefighting; proactively improve your AI, catch issues early, and eliminate black boxes.]]></description>
      <pubDate>Thu, 30 Oct 2025 00:00:00 GMT</pubDate>
      <author>Adham Sersour</author>
      <category>ai</category>
      <category>generative-ai</category>
      <enclosure url="/media/blog_images/ai_evalaution_monitoring_oUS3jsj.jpg" type="image/jpeg"/>
    </item>
    <item>
      <title><![CDATA[Evaluation-First LLMOps: Building Production-Grade Generative AI Applications]]></title>
      <link>https://sciences44.com/blog/evaluation-first-llmops-building-production-grade-generative-ai-applications</link>
      <guid isPermaLink="true">https://sciences44.com/blog/evaluation-first-llmops-building-production-grade-generative-ai-applications</guid>
      <description><![CDATA[Building GenAI apps isn't like traditional software – most projects fail without a crucial shift. Discover how a data-first, evaluation-centric approach unlocks success, and why you're likely chunking your data wrong. Stay ahead of the curve: this definitive guide is vital for any AI team scaling GenAI to production.]]></description>
      <pubDate>Tue, 09 Sep 2025 00:00:00 GMT</pubDate>
      <author>Adham Sersour</author>
      <category>ai</category>
      <category>generative-ai</category>
      <enclosure url="/media/blog_images/Neon_DNA_Helix_in_Cyberpunk_Glow.jpg" type="image/jpeg"/>
    </item>
    <item>
      <title><![CDATA[What is an AI Agent? A Simple Guide to Understanding the Technology]]></title>
      <link>https://sciences44.com/blog/what-is-an-ai-agent-a-simple-guide-to-understanding-the-technology</link>
      <guid isPermaLink="true">https://sciences44.com/blog/what-is-an-ai-agent-a-simple-guide-to-understanding-the-technology</guid>
      <description><![CDATA[LLMs only *talk*; AI agents *act* – orchestrating tools for real-world tasks. Even seasoned AI pros misunderstand agents' inner workings – this guide unveils the surprisingly simple core. Master key patterns like ReAct to build powerful, autonomous systems, but beware: production challenges are real.]]></description>
      <pubDate>Mon, 11 Aug 2025 00:00:00 GMT</pubDate>
      <author>Adham Sersour</author>
      <category>ai</category>
      <category>generative-ai</category>
      <enclosure url="/media/blog_images/gaussian_networks_agent.jpeg" type="image/jpeg"/>
    </item>
    <item>
      <title><![CDATA[🚀 Vector Databases: From Confusion to Clarity in Google Cloud's AI Ecosystem]]></title>
      <link>https://sciences44.com/blog/vector-databases-from-confusion-to-clarity-in-google-clouds-ai-ecosystem</link>
      <guid isPermaLink="true">https://sciences44.com/blog/vector-databases-from-confusion-to-clarity-in-google-clouds-ai-ecosystem</guid>
      <description><![CDATA[Google Cloud's AI services overwhelm? 85% of AI projects fail due to poor tech choices. Discover the PACE framework to avoid crippling choice paralysis and wasted weeks. Unlock your AI potential: choose the right vector database *before* your project becomes another statistic.]]></description>
      <pubDate>Wed, 30 Jul 2025 00:00:00 GMT</pubDate>
      <author>Adham Sersour</author>
      <category>ai</category>
      <category>generative-ai</category>
      <enclosure url="/static/images/vector_database/gcloud_solutions_vectordatabase.jpeg" type="image/jpeg"/>
    </item>
    <item>
      <title><![CDATA[🚀 Introduction to Vector Search]]></title>
      <link>https://sciences44.com/blog/introduction-to-vector-search</link>
      <guid isPermaLink="true">https://sciences44.com/blog/introduction-to-vector-search</guid>
      <description><![CDATA[Ditch keyword search! Vector databases unlock hidden meaning in your data, revealing connections you never knew existed. Discover how "embeddings" transform text, images, and more into searchable mathematical representations. Master vector search and build cutting-edge applications like personalized recommendations and AI-powered chatbots—the future of search is here.]]></description>
      <pubDate>Tue, 29 Jul 2025 00:00:00 GMT</pubDate>
      <author>Adham Sersour</author>
      <category>machine-learning</category>
      <enclosure url="/static/images/vector_database/embedding.jpg" type="image/jpeg"/>
    </item>
    <item>
      <title><![CDATA[PGVector on CloudSQL for GCP]]></title>
      <link>https://sciences44.com/blog/pgvector-on-cloudsql-for-gcp</link>
      <guid isPermaLink="true">https://sciences44.com/blog/pgvector-on-cloudsql-for-gcp</guid>
      <description><![CDATA[Automate PGVector on Cloud SQL deployment using Terraform and Makefiles! Deploy a complete vector search infrastructure with single commands; discover how. Build powerful semantic apps faster and easier, saving time and reducing errors.]]></description>
      <pubDate>Mon, 28 Jul 2025 00:00:00 GMT</pubDate>
      <author>Adham Sersour</author>
      <category>generative-ai</category>
      <enclosure url="/static/images/vector_database/pgvector_terraform.jpg" type="image/jpeg"/>
    </item>
  </channel>
</rss>