<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>NLP | Jay Polra</title><link>https://jaypolra.github.io/tags/nlp/</link><atom:link href="https://jaypolra.github.io/tags/nlp/index.xml" rel="self" type="application/rss+xml"/><description>NLP</description><generator>Hugo Blox Builder (https://hugoblox.com)</generator><language>en-us</language><lastBuildDate>Mon, 01 Jan 2024 00:00:00 +0000</lastBuildDate><image><url>https://jaypolra.github.io/media/icon_hu_3943bc2ea2cd9fa1.png</url><title>NLP</title><link>https://jaypolra.github.io/tags/nlp/</link></image><item><title>Candidate Matcher</title><link>https://jaypolra.github.io/project/candidate-matcher/</link><pubDate>Mon, 01 Jan 2024 00:00:00 +0000</pubDate><guid>https://jaypolra.github.io/project/candidate-matcher/</guid><description>&lt;p>Keyword overlap misses semantic alignment. A resume can match a job description on surface terms and still be irrelevant.&lt;/p>
&lt;p>Built ATS-style ranking using MiniLM embeddings and cosine similarity. LLM-generated summaries explain the match rather than just scoring it. Robust multi-format parsing pipeline for PDF, DOCX, and TXT files with fallback extraction logic. Deployed on Streamlit Cloud with Gemma 2B for summarization and Flan-T5 as a CPU-compatible fallback.&lt;/p></description></item></channel></rss>