<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: nravic</title><link>https://news.ycombinator.com/user?id=nravic</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Sat, 05 Sep 2026 05:39:44 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=nravic" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[Icons as a Service]]></title><description><![CDATA[
<p>Article URL: <a href="https://nravic.com/blog/images/">https://nravic.com/blog/images/</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49570597">https://news.ycombinator.com/item?id=49570597</a></p>
<p>Points: 34</p>
<p># Comments: 2</p>
]]></description><pubDate>Fri, 04 Sep 2026 21:44:59 +0000</pubDate><link>https://nravic.com/blog/images/</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=49570597</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49570597</guid></item><item><title><![CDATA[New comment by nravic in "Scamlexity: When agentic AI browsers get scammed"]]></title><description><![CDATA[
<p>I think even easier in fact - what's happening behind the scenes w/ an LLM is far more opaque</p>
]]></description><pubDate>Mon, 25 Aug 2025 15:00:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=45014612</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=45014612</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45014612</guid></item><item><title><![CDATA[New comment by nravic in "Compiling LLMs into a MegaKernel: A path to low-latency inference"]]></title><description><![CDATA[
<p>This is super interesting! We do something similar I think by taking a checkpoint after model initialization. I'm curious what you think about our approach, here's some benchmarks: <a href="https://docs.cedana.ai/articles/performance-of-cedanas-gpu-interception">https://docs.cedana.ai/articles/performance-of-cedanas-gpu-i...</a><p>We do some on-the-fly optimizations as well (like compiling into CUDA graphs or fusing together calls) which ends up resulting (for some inference engines) faster token throughput too.</p>
]]></description><pubDate>Fri, 20 Jun 2025 12:30:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=44327062</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=44327062</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=44327062</guid></item><item><title><![CDATA[New comment by nravic in "CRIU, a project to implement checkpoint/restore functionality for Linux"]]></title><description><![CDATA[
<p>How were you handling GPU state w/ pytorch? We added some custom code around CRIU to enable GPU checkpointing fwiw: <a href="https://docs.cedana.ai/setup/gpu-checkpointing/">https://docs.cedana.ai/setup/gpu-checkpointing/</a></p>
]]></description><pubDate>Sat, 22 Jun 2024 00:06:16 +0000</pubDate><link>https://news.ycombinator.com/item?id=40755100</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=40755100</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40755100</guid></item><item><title><![CDATA[New comment by nravic in "CRIU, a project to implement checkpoint/restore functionality for Linux"]]></title><description><![CDATA[
<p>We do this using CRIU right now! <a href="https://github.com/cedana/cedana">https://github.com/cedana/cedana</a><p>In fact one of our customer's use cases is exactly what you describe, allowing users to "hibernate" container workspaces.</p>
]]></description><pubDate>Fri, 21 Jun 2024 20:54:20 +0000</pubDate><link>https://news.ycombinator.com/item?id=40753702</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=40753702</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40753702</guid></item><item><title><![CDATA[New comment by nravic in "CRIU, a project to implement checkpoint/restore functionality for Linux"]]></title><description><![CDATA[
<p>I'm probably the cofounder of the guy you spoke with! Here's our repo: <a href="https://github.com/cedana/cedana">https://github.com/cedana/cedana</a></p>
]]></description><pubDate>Fri, 21 Jun 2024 20:51:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=40753680</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=40753680</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=40753680</guid></item><item><title><![CDATA[New comment by nravic in "280M e-bikes and mopeds are cutting demand for oil far more than electric cars"]]></title><description><![CDATA[
<p>what kind of pannier bags do you use to haul groceries? All the ones I've used have been too small for my purposes.</p>
]]></description><pubDate>Fri, 17 Nov 2023 16:37:53 +0000</pubDate><link>https://news.ycombinator.com/item?id=38305917</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=38305917</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=38305917</guid></item><item><title><![CDATA[New comment by nravic in "Kullback–Leibler divergence"]]></title><description><![CDATA[
<p>IIRC (and in my experience) KL divergence doesn't account for double counting. Wrote a paper where I ended up having to use a custom metric instead: 
<a href="https://digitalcommons.usu.edu/cgi/viewcontent.cgi?article=4438&context=smallsat" rel="nofollow noreferrer">https://digitalcommons.usu.edu/cgi/viewcontent.cgi?article=4...</a></p>
]]></description><pubDate>Tue, 22 Aug 2023 18:26:17 +0000</pubDate><link>https://news.ycombinator.com/item?id=37226709</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=37226709</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=37226709</guid></item><item><title><![CDATA[New comment by nravic in "Show HN: Rivet – Open-source game server management with Nomad and Rust"]]></title><description><![CDATA[
<p>This is so cool! Out of curiosity, could you walk through your decisionmaking around using Nix and how it fits in with Rivet?</p>
]]></description><pubDate>Sat, 19 Aug 2023 15:30:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=37189756</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=37189756</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=37189756</guid></item><item><title><![CDATA[New comment by nravic in "Show HN: Retake – Open-Source Hybrid Search for Postgres"]]></title><description><![CDATA[
<p>how often are these batch jobs run? I'm curious to know what the absolute maximum sync frequency can be.</p>
]]></description><pubDate>Thu, 10 Aug 2023 17:34:28 +0000</pubDate><link>https://news.ycombinator.com/item?id=37079124</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=37079124</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=37079124</guid></item><item><title><![CDATA[New comment by nravic in "Guide to running Llama 2 locally"]]></title><description><![CDATA[
<p>Self plug: run llama.cpp as an inference server on a spot instance anywhere: <a href="https://cedana.readthedocs.io/en/latest/examples.html#running-llama-cpp-inference" rel="nofollow noreferrer">https://cedana.readthedocs.io/en/latest/examples.html#runnin...</a></p>
]]></description><pubDate>Tue, 25 Jul 2023 23:56:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=36871354</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=36871354</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=36871354</guid></item><item><title><![CDATA[New comment by nravic in "What are some interesting uses of WebUSB you have come across?"]]></title><description><![CDATA[
<p>GrapheneOS uses WebUSB to flash your phone.<p>Pretty wild first experience doing it for me - I'm used to the hassle of sideloading a bootloader and then flashing.</p>
]]></description><pubDate>Sun, 29 Jan 2023 01:06:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=34563830</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=34563830</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=34563830</guid></item><item><title><![CDATA[New comment by nravic in "Ask HN: Who is hiring? (July 2022)"]]></title><description><![CDATA[
<p>Hey! I'd love to learn more about the position - the hook @ jpl email on your website is bouncing though. Do you mind sending me contact details so we can chat?<p>Thanks :)</p>
]]></description><pubDate>Thu, 07 Jul 2022 14:13:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=32014363</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=32014363</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=32014363</guid></item><item><title><![CDATA[New comment by nravic in "This career feels like a few key hours with year-long cool-down periods"]]></title><description><![CDATA[
<p>Do people out of college even do Leetcode anymore? Surely there's a better use of time once you're in a paying job.</p>
]]></description><pubDate>Sat, 02 Oct 2021 11:48:23 +0000</pubDate><link>https://news.ycombinator.com/item?id=28727833</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=28727833</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=28727833</guid></item><item><title><![CDATA[New comment by nravic in "The death of corporate research labs"]]></title><description><![CDATA[
<p>If you're looking at aerospace/defense startups, this is usually the case.<p>It's not unheard of for them to employ people who's primary role is grant writing to try and get (for example) SBIR funding</p>
]]></description><pubDate>Thu, 21 May 2020 14:45:34 +0000</pubDate><link>https://news.ycombinator.com/item?id=23258967</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=23258967</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=23258967</guid></item><item><title><![CDATA[New comment by nravic in "Ask HN: What's your cross-platform PDF / ePub reading workflow?"]]></title><description><![CDATA[
<p>Zotero is my go-to</p>
]]></description><pubDate>Tue, 28 Jan 2020 16:56:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=22171208</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=22171208</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=22171208</guid></item><item><title><![CDATA[New comment by nravic in "Clojure – Fascinated, Disappointed, Astonished (2016)"]]></title><description><![CDATA[
<p>Julia's REPL is beautiful</p>
]]></description><pubDate>Fri, 24 Jan 2020 19:05:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=22141363</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=22141363</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=22141363</guid></item><item><title><![CDATA[New comment by nravic in "Computational Power Found in the Arms of Neurons"]]></title><description><![CDATA[
<p>Yeah, spiking neural networks are a tough nut to crack. Check out Nengo if you're interested in learning more</p>
]]></description><pubDate>Thu, 16 Jan 2020 16:08:52 +0000</pubDate><link>https://news.ycombinator.com/item?id=22065818</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=22065818</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=22065818</guid></item><item><title><![CDATA[New comment by nravic in "Show HN: Write a private diary using good old email"]]></title><description><![CDATA[
<p>What are Protonmail's problems?</p>
]]></description><pubDate>Tue, 14 Jan 2020 18:19:39 +0000</pubDate><link>https://news.ycombinator.com/item?id=22047615</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=22047615</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=22047615</guid></item><item><title><![CDATA[New comment by nravic in "China's growing threat to academic freedom"]]></title><description><![CDATA[
<p>very visceral reaction there</p>
]]></description><pubDate>Wed, 27 Nov 2019 17:24:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=21649851</link><dc:creator>nravic</dc:creator><comments>https://news.ycombinator.com/item?id=21649851</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=21649851</guid></item></channel></rss>