<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
    <channel>
        <title>Dhruvkumar Patel | Data Scientist &amp; ML Engineer</title>
        <link>https://www.stackdhruv.com/</link>
        <description>Dhruvkumar Patel, Data Scientist at Myntra. M.Tech CSE from IIIT Delhi, building production ads ranking systems.</description>
        <lastBuildDate>Thu, 01 Oct 2026 21:36:49 GMT</lastBuildDate>
        <docs>https://validator.w3.org/feed/docs/rss2.html</docs>
        <generator>Feed for Node.js</generator>
        <copyright>All rights reserved 2026, Dhruvkumar Patel | Data Scientist &amp; ML Engineer</copyright>
        <atom:link href="https://www.stackdhruv.com/rss.xml" rel="self" type="application/rss+xml"/>
        <item>
            <title><![CDATA[Jev vs LAYA: Which Should Replace Your LLM Call?]]></title>
            <link>https://www.stackdhruv.com/blog/jev-vs-laya-llm-classifier</link>
            <guid>https://www.stackdhruv.com/blog/jev-vs-laya-llm-classifier</guid>
            <pubDate>Thu, 01 Oct 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[Jev and LAYA launched in the same week of September 2026 to replace the LLM call that returns one word, and most launch-week comparisons measured different things. Checked against independent benchmarks and LAYA's own code, their confidence scores hold only on inputs like the training data, and a ModernBERT classifier trained on about 2,000 labels scored above hosted Jev on a CPU.]]></description>
            <category>LLM Classification</category>
            <category>Calibration</category>
            <category>TypeSafe JEV</category>
            <category>LAYA</category>
            <category>ModernBERT</category>
            <category>Zero Short Classification</category>
            <category>Text Classification</category>
        </item>
        <item>
            <title><![CDATA[Mental Health Meme Classification]]></title>
            <link>https://www.stackdhruv.com/blog/mental-health-meme-classification</link>
            <guid>https://www.stackdhruv.com/blog/mental-health-meme-classification</guid>
            <pubDate>Mon, 15 Jun 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[A multimodal system for classifying depression and anxiety symptoms from internet memes, combining OCR text, figurative reasoning generated by a vision-language model, and visual features. What the measurements show is that the figurative reasoning pathway is where nearly all of the gain comes from, and what they leave open is whether the quality of that reasoning is consistent enough to build on.]]></description>
            <category>NLP</category>
            <category>Multimodal AI</category>
            <category>PyTorch</category>
            <category>MentalBART</category>
            <category>Qwen2.5-VL</category>
            <category>Streamlit</category>
        </item>
        <item>
            <title><![CDATA[Microservices Benchmarking with DeathStarBench]]></title>
            <link>https://www.stackdhruv.com/blog/microservices-benchmarking-with-deathstarbench</link>
            <guid>https://www.stackdhruv.com/blog/microservices-benchmarking-with-deathstarbench</guid>
            <pubDate>Sun, 24 May 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[A course project deploying DeathStarBench's Social Network across Docker Compose, local k3s, and GKE to benchmark performance under different replica and node configurations. Getting the system observable turned out to be a harder problem than benchmarking it, and once we could see inside the cluster, the component behaving oddly was the load balancer rather than the application.]]></description>
            <category>Docker</category>
            <category>Kubernetes</category>
            <category>GKE</category>
            <category>Helm</category>
            <category>Pixie</category>
            <category>Jaeger</category>
            <category>wrk2</category>
            <category>Observability</category>
            <category>Microservices</category>
        </item>
        <item>
            <title><![CDATA[Advanced ANPR & Face Recognition]]></title>
            <link>https://www.stackdhruv.com/blog/advanced-anpr-face-recognition</link>
            <guid>https://www.stackdhruv.com/blog/advanced-anpr-face-recognition</guid>
            <pubDate>Wed, 21 May 2025 00:00:00 GMT</pubDate>
            <description><![CDATA[An end-to-end traffic surveillance system pairing YOLOv8 license plate detection with a Siamese few-shot face recognizer to cross-verify driver identity against vehicle registration ,  trained on real CCTV footage obtained under NDA from the Ahmedabad West traffic police, and a national finalist at the KAVACH-2023 hackathon. The core insight: a plate identifies a vehicle, but enforcement requires knowing who is driving it, and the data you can collect on drivers is nothing like the data you can collect on plates.]]></description>
            <category>Python</category>
            <category>PyTorch</category>
            <category>YOLOv8</category>
            <category>ResNet</category>
            <category>OpenCV</category>
            <category>Android</category>
            <category>Kotlin</category>
            <category>Flask</category>
            <category>Computer Vision</category>
            <category>ANPR</category>
        </item>
        <item>
            <title><![CDATA[Student Dropout Analysis]]></title>
            <link>https://www.stackdhruv.com/blog/student-dropout-analysis</link>
            <guid>https://www.stackdhruv.com/blog/student-dropout-analysis</guid>
            <pubDate>Mon, 10 Jun 2024 00:00:00 GMT</pubDate>
            <description><![CDATA[A two-sided platform, EduDropX, pairing a Flask school administration portal with a React government analytics dashboard, built to turn UDISE+ dropout data from an aggregate reporting artifact into a system schools and officials can actually act on. Published at IEEE I2CT 2024, and the honest limitation is that the machine learning model predicts dropout rates by category, not risk for an individual student.]]></description>
            <category>Python</category>
            <category>Machine Learning</category>
            <category>Data Visualization</category>
            <category>Hackathon</category>
            <category>IEEE</category>
            <category>React</category>
            <category>Flask</category>
        </item>
    </channel>
</rss>