<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Projects on Deepak Singh</title><link>https://deepak91168.github.io/portfolio/projects/</link><description>Recent content in Projects on Deepak Singh</description><generator>Hugo</generator><language>en</language><lastBuildDate>Thu, 16 Apr 2026 00:00:00 +0530</lastBuildDate><atom:link href="https://deepak91168.github.io/portfolio/projects/index.xml" rel="self" type="application/rss+xml"/><item><title>B.Tech Major Project: Legal Judgment Summarization using Transformer Models</title><link>https://deepak91168.github.io/portfolio/projects/btech-major-project-nit-hamirpur/</link><pubDate>Thu, 16 Apr 2026 00:00:00 +0530</pubDate><guid>https://deepak91168.github.io/portfolio/projects/btech-major-project-nit-hamirpur/</guid><description>&lt;h2 id="overview"&gt;Overview&lt;/h2&gt;
&lt;p&gt;Designed and implemented an end-to-end NLP system to automatically summarize long-form Indian court judgments using transformer-based architectures.&lt;/p&gt;
&lt;p&gt;This project focuses on handling &lt;strong&gt;long-context documents&lt;/strong&gt;, &lt;strong&gt;domain-specific language&lt;/strong&gt;, and &lt;strong&gt;evaluation of abstractive summarization models&lt;/strong&gt;.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="problem-statement"&gt;Problem Statement&lt;/h2&gt;
&lt;p&gt;Legal judgments are often lengthy and complex, making them difficult to quickly interpret.&lt;br&gt;
The goal was to generate concise, meaningful summaries while preserving legal semantics.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="tech-stack"&gt;Tech Stack&lt;/h2&gt;
&lt;p&gt;Python, HuggingFace Transformers, PyTorch, NLP&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="system-design"&gt;System Design&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;&lt;strong&gt;Data Preprocessing&lt;/strong&gt;&lt;/p&gt;</description></item><item><title>Stdlib: High-Precision Mathematical Functions (Open Source Contribution)</title><link>https://deepak91168.github.io/portfolio/projects/stdlib-opensource-contributions/</link><pubDate>Thu, 16 Apr 2026 00:00:00 +0530</pubDate><guid>https://deepak91168.github.io/portfolio/projects/stdlib-opensource-contributions/</guid><description>&lt;h2 id="overview"&gt;Overview&lt;/h2&gt;
&lt;p&gt;Contributed to stdlib, a scientific computing library, by implementing high-performance and IEEE-compliant mathematical functions in both C and JavaScript.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="motivation"&gt;Motivation&lt;/h2&gt;
&lt;p&gt;Accurate numerical computation is critical in scientific and engineering applications.&lt;br&gt;
The goal was to implement functions with &lt;strong&gt;high precision&lt;/strong&gt;, &lt;strong&gt;performance&lt;/strong&gt;, and &lt;strong&gt;cross-language compatibility&lt;/strong&gt;.&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="tech-stack"&gt;Tech Stack&lt;/h2&gt;
&lt;p&gt;C, JavaScript, Node.js&lt;/p&gt;
&lt;hr&gt;
&lt;h2 id="work-done"&gt;Work Done&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;Implemented:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;code&gt;exp10f&lt;/code&gt; (base-10 exponential function)&lt;/li&gt;
&lt;li&gt;&lt;code&gt;csignumf&lt;/code&gt; with &lt;code&gt;.assign&lt;/code&gt; and &lt;code&gt;.strided&lt;/code&gt; variants&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;Built complete packages including:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Native C implementations&lt;/li&gt;
&lt;li&gt;JavaScript bindings&lt;/li&gt;
&lt;li&gt;Benchmark suites&lt;/li&gt;
&lt;li&gt;Unit tests&lt;/li&gt;
&lt;/ul&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;Followed &lt;strong&gt;IEEE-754 floating-point standards&lt;/strong&gt;&lt;/p&gt;</description></item></channel></rss>