<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Deep Learning | Shiv Kumar | Lead Engineer – GenAI &amp; Agentic AI</title><link>https://ms-kumar.github.io/shivkumar-portfolio/tags/deep-learning/</link><atom:link href="https://ms-kumar.github.io/shivkumar-portfolio/tags/deep-learning/index.xml" rel="self" type="application/rss+xml"/><description>Deep Learning</description><generator>HugoBlox Kit (https://hugoblox.com)</generator><language>en-us</language><lastBuildDate>Wed, 01 Nov 2023 00:00:00 +0000</lastBuildDate><image><url>https://ms-kumar.github.io/shivkumar-portfolio/media/icon_hu_1c0e9cb08cfb822a.png</url><title>Deep Learning</title><link>https://ms-kumar.github.io/shivkumar-portfolio/tags/deep-learning/</link></image><item><title>ClothFormer: A Boundary-Aware Self-Attention Network for Human Outfit Parsing</title><link>https://ms-kumar.github.io/shivkumar-portfolio/publications/clothformer-icrais-2023/</link><pubDate>Wed, 01 Nov 2023 00:00:00 +0000</pubDate><guid>https://ms-kumar.github.io/shivkumar-portfolio/publications/clothformer-icrais-2023/</guid><description>&lt;p&gt;Read the paper on
.&lt;/p&gt;</description></item><item><title>DCSP-RCNN: A Network for Person Detection with Color, Size and Pattern Characteristic Parsing</title><link>https://ms-kumar.github.io/shivkumar-portfolio/publications/dcsp-rcnn-icrais-2023/</link><pubDate>Wed, 01 Nov 2023 00:00:00 +0000</pubDate><guid>https://ms-kumar.github.io/shivkumar-portfolio/publications/dcsp-rcnn-icrais-2023/</guid><description>&lt;p&gt;Read the paper on
.&lt;/p&gt;</description></item><item><title>Deep Learning Kernels — Accelerator Optimization</title><link>https://ms-kumar.github.io/shivkumar-portfolio/projects/dl-kernels/</link><pubDate>Fri, 01 Jan 2021 00:00:00 +0000</pubDate><guid>https://ms-kumar.github.io/shivkumar-portfolio/projects/dl-kernels/</guid><description>&lt;h2 id="project-details"&gt;Project Details&lt;/h2&gt;
&lt;div class="not-prose my-6 grid grid-cols-1 gap-4"&gt;
&lt;div class="group rounded-2xl border border-gray-200 bg-white/70 p-4 shadow-sm transition-all hover:-translate-y-0.5 hover:border-primary-300 hover:shadow-md dark:border-white/[0.08] dark:bg-white/[0.03] dark:hover:border-primary-500/40"&gt;
&lt;div class="flex items-start gap-2.5"&gt;
&lt;div class="mt-0.5 flex h-7 w-7 shrink-0 items-center justify-center rounded-lg bg-primary-600 text-[11px] font-bold text-white shadow-sm shadow-primary-500/25 dark:bg-primary-500"&gt;
01
&lt;/div&gt;
&lt;div&gt;
&lt;h3 class="m-0 text-base font-semibold text-gray-900 dark:text-white"&gt;Accelerator Framework&lt;/h3&gt;
&lt;div class="mt-1 text-sm leading-5 text-gray-600 dark:text-gray-300"&gt;Developed a PyTorch and TensorFlow software suite for high-performance accelerator training and inference.&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;div class="group rounded-2xl border border-gray-200 bg-white/70 p-4 shadow-sm transition-all hover:-translate-y-0.5 hover:border-primary-300 hover:shadow-md dark:border-white/[0.08] dark:bg-white/[0.03] dark:hover:border-primary-500/40"&gt;
&lt;div class="flex items-start gap-2.5"&gt;
&lt;div class="mt-0.5 flex h-7 w-7 shrink-0 items-center justify-center rounded-lg bg-primary-600 text-[11px] font-bold text-white shadow-sm shadow-primary-500/25 dark:bg-primary-500"&gt;
02
&lt;/div&gt;
&lt;div&gt;
&lt;h3 class="m-0 text-base font-semibold text-gray-900 dark:text-white"&gt;Kernel Debugging&lt;/h3&gt;
&lt;div class="mt-1 text-sm leading-5 text-gray-600 dark:text-gray-300"&gt;Analyzed and resolved kernel bugs and performance bottlenecks.&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;div class="group rounded-2xl border border-gray-200 bg-white/70 p-4 shadow-sm transition-all hover:-translate-y-0.5 hover:border-primary-300 hover:shadow-md dark:border-white/[0.08] dark:bg-white/[0.03] dark:hover:border-primary-500/40"&gt;
&lt;div class="flex items-start gap-2.5"&gt;
&lt;div class="mt-0.5 flex h-7 w-7 shrink-0 items-center justify-center rounded-lg bg-primary-600 text-[11px] font-bold text-white shadow-sm shadow-primary-500/25 dark:bg-primary-500"&gt;
03
&lt;/div&gt;
&lt;div&gt;
&lt;h3 class="m-0 text-base font-semibold text-gray-900 dark:text-white"&gt;TPC Mapping&lt;/h3&gt;
&lt;div class="mt-1 text-sm leading-5 text-gray-600 dark:text-gray-300"&gt;Mapped PyTorch kernels to TPC kernels in C++.&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;div class="group rounded-2xl border border-gray-200 bg-white/70 p-4 shadow-sm transition-all hover:-translate-y-0.5 hover:border-primary-300 hover:shadow-md dark:border-white/[0.08] dark:bg-white/[0.03] dark:hover:border-primary-500/40"&gt;
&lt;div class="flex items-start gap-2.5"&gt;
&lt;div class="mt-0.5 flex h-7 w-7 shrink-0 items-center justify-center rounded-lg bg-primary-600 text-[11px] font-bold text-white shadow-sm shadow-primary-500/25 dark:bg-primary-500"&gt;
04
&lt;/div&gt;
&lt;div&gt;
&lt;h3 class="m-0 text-base font-semibold text-gray-900 dark:text-white"&gt;Custom Operators&lt;/h3&gt;
&lt;div class="mt-1 text-sm leading-5 text-gray-600 dark:text-gray-300"&gt;Developed operators using combinations of TPC kernels.&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;div class="group rounded-2xl border border-gray-200 bg-white/70 p-4 shadow-sm transition-all hover:-translate-y-0.5 hover:border-primary-300 hover:shadow-md dark:border-white/[0.08] dark:bg-white/[0.03] dark:hover:border-primary-500/40"&gt;
&lt;div class="flex items-start gap-2.5"&gt;
&lt;div class="mt-0.5 flex h-7 w-7 shrink-0 items-center justify-center rounded-lg bg-primary-600 text-[11px] font-bold text-white shadow-sm shadow-primary-500/25 dark:bg-primary-500"&gt;
05
&lt;/div&gt;
&lt;div&gt;
&lt;h3 class="m-0 text-base font-semibold text-gray-900 dark:text-white"&gt;Model Export&lt;/h3&gt;
&lt;div class="mt-1 text-sm leading-5 text-gray-600 dark:text-gray-300"&gt;Converted deep learning models to ONNX for deployment.&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;div class="group rounded-2xl border border-gray-200 bg-white/70 p-4 shadow-sm transition-all hover:-translate-y-0.5 hover:border-primary-300 hover:shadow-md dark:border-white/[0.08] dark:bg-white/[0.03] dark:hover:border-primary-500/40"&gt;
&lt;div class="flex items-start gap-2.5"&gt;
&lt;div class="mt-0.5 flex h-7 w-7 shrink-0 items-center justify-center rounded-lg bg-primary-600 text-[11px] font-bold text-white shadow-sm shadow-primary-500/25 dark:bg-primary-500"&gt;
06
&lt;/div&gt;
&lt;div&gt;
&lt;h3 class="m-0 text-base font-semibold text-gray-900 dark:text-white"&gt;Graph Validation&lt;/h3&gt;
&lt;div class="mt-1 text-sm leading-5 text-gray-600 dark:text-gray-300"&gt;Used Netron to visualize and validate exported model graphs.&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;div class="group rounded-2xl border border-gray-200 bg-white/70 p-4 shadow-sm transition-all hover:-translate-y-0.5 hover:border-primary-300 hover:shadow-md dark:border-white/[0.08] dark:bg-white/[0.03] dark:hover:border-primary-500/40"&gt;
&lt;div class="flex items-start gap-2.5"&gt;
&lt;div class="mt-0.5 flex h-7 w-7 shrink-0 items-center justify-center rounded-lg bg-primary-600 text-[11px] font-bold text-white shadow-sm shadow-primary-500/25 dark:bg-primary-500"&gt;
07
&lt;/div&gt;
&lt;div&gt;
&lt;h3 class="m-0 text-base font-semibold text-gray-900 dark:text-white"&gt;Performance Focus&lt;/h3&gt;
&lt;div class="mt-1 text-sm leading-5 text-gray-600 dark:text-gray-300"&gt;Optimized execution across accelerator training and inference workloads.&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;h2 id="tech-stack"&gt;Tech Stack&lt;/h2&gt;
&lt;div class="not-prose my-6 grid grid-cols-1 gap-4"&gt;
&lt;div class="rounded-xl border border-gray-200 bg-white/70 p-4 shadow-sm transition-colors hover:border-primary-300 dark:border-white/[0.08] dark:bg-white/[0.03] dark:hover:border-primary-500/40"&gt;
&lt;p class="m-0 text-xs font-semibold uppercase tracking-[0.16em] text-gray-500 dark:text-gray-400"&gt;Frameworks &amp;amp; Data&lt;/p&gt;
&lt;div class="mt-3 flex flex-wrap gap-2"&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;Python&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;NumPy&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;Pandas&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;PyTorch&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;TensorFlow&lt;/span&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;div class="rounded-xl border border-gray-200 bg-white/70 p-4 shadow-sm transition-colors hover:border-primary-300 dark:border-white/[0.08] dark:bg-white/[0.03] dark:hover:border-primary-500/40"&gt;
&lt;p class="m-0 text-xs font-semibold uppercase tracking-[0.16em] text-gray-500 dark:text-gray-400"&gt;Acceleration &amp;amp; Tooling&lt;/p&gt;
&lt;div class="mt-3 flex flex-wrap gap-2"&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;C&amp;#43;&amp;#43;&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;TPC Kernels&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;ONNX&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;Netron&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;VS Code&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;Linux&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;Git&lt;/span&gt;
&lt;span class="rounded-full border border-primary-200 bg-primary-50 px-2.5 py-1 text-xs font-medium text-primary-700 dark:border-primary-500/20 dark:bg-primary-500/10 dark:text-primary-300"&gt;JIRA&lt;/span&gt;
&lt;/div&gt;
&lt;/div&gt;
&lt;/div&gt;</description></item></channel></rss>