<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: benlm</title><link>https://news.ycombinator.com/user?id=benlm</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Thu, 19 Nov 2026 21:42:31 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=benlm" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by benlm in "Bringing Up DeepSeek-V4-Flash on AMD MI300X"]]></title><description><![CDATA[
<p>Nice work! Would DeepSeek V4 Pro on 8xMI300X work with these patches?</p>
]]></description><pubDate>Tue, 02 Jun 2026 19:23:19 +0000</pubDate><link>https://news.ycombinator.com/item?id=48374964</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=48374964</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48374964</guid></item><item><title><![CDATA[New comment by benlm in "We replaced RAG with a virtual filesystem for our AI documentation assistant"]]></title><description><![CDATA[
<p>Our vfs is also pretty powerful too, though: it is all backed by postgres then projected into files/directories for our agents. They get basic grep etc but also optimized fts tools for bm25, jq, and preview tools that show representative slices of large documents. All on top of Pydantic AI.</p>
]]></description><pubDate>Sat, 04 Apr 2026 14:34:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=47639412</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=47639412</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47639412</guid></item><item><title><![CDATA[New comment by benlm in "We replaced RAG with a virtual filesystem for our AI documentation assistant"]]></title><description><![CDATA[
<p>We use both a virtual file system and RAG — they each excel in different areas. The trick with RAG is the quality of data: we use an LLM to chunk into semantically cohesive sections, as well as generate metadata (including fact triples and links to other related chunks in the document) for every chunk as well as the document as a whole. We use voyage contextual embeddings to then embed each chunk with the document and chunk metadata. Works incredibly well. At retrieval time the agent can follow chunk links if needed, as well as analyze the raw file in the vfs. High quality instruction based reranking helps a lot too! We are often looking over 10s of thousands of documents and it’d be very inefficient to have our agents analyze just the vfs without rag.</p>
]]></description><pubDate>Sat, 04 Apr 2026 14:31:00 +0000</pubDate><link>https://news.ycombinator.com/item?id=47639382</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=47639382</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47639382</guid></item><item><title><![CDATA[New comment by benlm in "Ask HN: Who is hiring? (November 2025)"]]></title><description><![CDATA[
<p>ShiftUp | Staff Software Engineer | Remote | <a href="https://shiftupai.com" rel="nofollow">https://shiftupai.com</a><p>Hiring for a full-time, 100% remote role (entire company is remote) for a Staff Software Engineer, working on the ShiftUp platform stack (Python, Pydantic AI, GCP, Temporal Workflows, Turbopuffer, Postgres). You will be one of the first five engineering hires, and you will have the ability to grow and shape your own role.<p>We need someone with solid experience designing and implementing systems with many moving parts, who can easily shift between systems design, writing and reviewing code (or, let's face it, supervising Codex/Cursor/Claude Code), working with the team to design features, and talking to customers to gain feedback and insights.<p>You'll be working to build all kinds of agents, workflows, and other components to support our features. Our research agents are truly deep, sometimes spending upwards of an hour searching and analyzing information for a single prospect or strategy. Our RAG pipeline is focused on quality, leveraging LLM chunking, metadata and graph generation, and contextual embeddings.<p>If this sounds like fun, let's chat: you won't be asked to regurgitate leetcode solutions, we'll simply have a conversation and see if there's a mutual fit. Our interview process is lightweight, respectful and moves fast. Competitive salary, equity and full benefits.<p>You can email me directly: ben@shiftupai.com<p>---<p>About the company: ShiftUp is a new kind of intelligence-driven sales platform that identifies emerging demand and drives revenue. Built from scratch to leverage LLM-driven agentic workflows, ShiftUp deeply researches prospects, identifies the highest-value opportunities, and provides targeted sales strategies to execute on them.<p>ShiftUp is delivered as a native app on Salesforce AppExchange, providing our platform to sellers where they already are. Our Salesforce app talks to a cloud-native stack that uses Python services, durable workflows, and SQL + vector databases.<p>We're well-funded, having recently closed a $3M seed round. We're building out a small team of highly talented, well-rounded engineers to deliver on our vision. Our company is 100% remote, and we use SoWork to help build our culture. As a small startup, everyone is highly motivated and focused on moving quickly, but we also place importance on a sustainable work-life balance so that you can continue to do your best work.</p>
]]></description><pubDate>Tue, 04 Nov 2025 19:06:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=45814715</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=45814715</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=45814715</guid></item><item><title><![CDATA[New comment by benlm in "Zed: High-performance AI Code Editor"]]></title><description><![CDATA[
<p>I really miss the everything is editable panel, it felt like a superpower. There’s a bit of a learning curve, but after it’s amazing and everything else feels limited.</p>
]]></description><pubDate>Wed, 07 May 2025 13:52:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=43915709</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=43915709</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=43915709</guid></item><item><title><![CDATA[New comment by benlm in "Show HN: Jumprun – AI-powered research as interactive canvases"]]></title><description><![CDATA[
<p>If anyone was trying it out, we just had an issue with one of our backend services hitting a usage limit. All fixed now so if you had a failed refresh please try again.</p>
]]></description><pubDate>Sun, 24 Mar 2024 18:05:59 +0000</pubDate><link>https://news.ycombinator.com/item?id=39808958</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=39808958</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39808958</guid></item><item><title><![CDATA[New comment by benlm in "Show HN: Jumprun – AI-powered research as interactive canvases"]]></title><description><![CDATA[
<p>OP here - feel free to ask any questions. We are still trying to figure out what direction to take Jumprun in from here and would love your feedback and ideas. Thanks!</p>
]]></description><pubDate>Sun, 24 Mar 2024 17:10:40 +0000</pubDate><link>https://news.ycombinator.com/item?id=39808514</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=39808514</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39808514</guid></item><item><title><![CDATA[Show HN: Jumprun – AI-powered research as interactive canvases]]></title><description><![CDATA[
<p>Hey HN! We're super excited to share our labor of love, Jumprun, which we've just opened up to everyone to try for free.<p>Jumprun is an AI-powered research tool that whips up stunning, interactive canvases. It can dive into websites, YouTube, X, and more. You can keep your canvases fresh with scheduled daily updates, and share them with the world.<p>Check out some cool canvases we've whipped up:<p>- Gleam language news: <a href="https://jumprun.ai/share/canvas/01HSM7G0TBYTJWDYRFET16DFJQ" rel="nofollow">https://jumprun.ai/share/canvas/01HSM7G0TBYTJWDYRFET16DFJQ</a>
- Cybertruck reviews: <a href="https://jumprun.ai/share/canvas/01HSKTGQEGB58Y1BABY0KRRAQ0" rel="nofollow">https://jumprun.ai/share/canvas/01HSKTGQEGB58Y1BABY0KRRAQ0</a>
- Apple Vision Pro reviews: <a href="https://jumprun.ai/share/canvas/01HNXB2K3GM7KPRP45Y2CVVJSC" rel="nofollow">https://jumprun.ai/share/canvas/01HNXB2K3GM7KPRP45Y2CVVJSC</a>
- Things to do in San Francisco: <a href="https://jumprun.ai/share/canvas/01HSPRS3YXXJ9TV90P18MBZRPS" rel="nofollow">https://jumprun.ai/share/canvas/01HSPRS3YXXJ9TV90P18MBZRPS</a><p>Jumprun was in closed beta for a bit because GPT4-Turbo was burning a hole in our pockets. But when Anthropic dropped Claude3 Haiku and Sonnet, we saw an opportunity to use them to slash costs and get Jumprun out to all of you. We're now rocking Haiku for research and Sonnet for the final "analysis," and the results are pretty great. We even think they might be better than GPT4-Turbo!<p>If you've got any technical questions, fire away! We're currently trying to nail down product-market fit and figure out where to focus re: features and use cases. We'd love to hear your thoughts!<p>Thanks a ton!
- Ben and Rico</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=39804337">https://news.ycombinator.com/item?id=39804337</a></p>
<p>Points: 35</p>
<p># Comments: 9</p>
]]></description><pubDate>Sun, 24 Mar 2024 01:20:28 +0000</pubDate><link>https://jumprun.ai/</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=39804337</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39804337</guid></item><item><title><![CDATA[New comment by benlm in "Show HN: I made an Apple Vision Pro review roundup with AI using Jumprun"]]></title><description><![CDATA[
<p>Here you go, AVP vs Quest 3: <a href="https://jumprun.ai/share/canvas/01HNXG4VK47Y0J1CC403Z6JSM0" rel="nofollow">https://jumprun.ai/share/canvas/01HNXG4VK47Y0J1CC403Z6JSM0</a></p>
]]></description><pubDate>Mon, 05 Feb 2024 21:26:23 +0000</pubDate><link>https://news.ycombinator.com/item?id=39267428</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=39267428</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39267428</guid></item><item><title><![CDATA[New comment by benlm in "Ask HN: What have you built with LLMs?"]]></title><description><![CDATA[
<p>We built Jumprun. You can use it to research and analyze data sources, and it'll produce beautiful canvases with tables, charts, videos, maps, etc. We're working on automations so you can setup natural language trigger conditions that execute actions.<p>We built it in Kotlin with Ktor server, htmx and tailwind. It uses a mixture of models, including gpt4-turbo, gpt4-vision and gemini-pro-vision. It's deployed using Kamal on bare metal.<p>Example canvas that provides a roundup of Apple Vision Pro reviews: <a href="https://jumprun.ai/share/canvas/01HNXB2K3GM7KPRP45Y2CVVJSC" rel="nofollow">https://jumprun.ai/share/canvas/01HNXB2K3GM7KPRP45Y2CVVJSC</a><p>Our learn more page with some screenshots to show creating a canvas: <a href="https://jumprun.ai/learn-more" rel="nofollow">https://jumprun.ai/learn-more</a><p>It's a free closed beta at the moment to control costs, but let me know if you'd like an invite.</p>
]]></description><pubDate>Mon, 05 Feb 2024 20:56:33 +0000</pubDate><link>https://news.ycombinator.com/item?id=39267040</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=39267040</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39267040</guid></item><item><title><![CDATA[New comment by benlm in "Show HN: I made an Apple Vision Pro review roundup with AI using Jumprun"]]></title><description><![CDATA[
<p>Very easy! You just need to ground it with the right data sources. Eg. you can add a web search datasource for "oculus reviews", and then add a table component to your canvas, instructing it to provide a comparative analysis.<p>The source material doesn't need to provide a comparative analysis itself: as long as it finds data detailing Vision Pro specs (which it already found) and Oculus specs it'll be able to do the analysis.<p>Canvases also have memory and auto-refresh, so once it finds the specs it can remember them across refreshes/updates, but also incorporate any new information it finds.</p>
]]></description><pubDate>Mon, 05 Feb 2024 20:17:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=39266517</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=39266517</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39266517</guid></item><item><title><![CDATA[New comment by benlm in "Show HN: I made an Apple Vision Pro review roundup with AI using Jumprun"]]></title><description><![CDATA[
<p>Just a little more context: Jumprun let's you connect different data sources (like web searches/pages, APIs, X, Youtube videos, Notion etc) and use LLMs to analyze and visualize the data.<p>We support rich components like tables, timeseries, charts and maps. We're working on automations at the moment so that you can provide natural language conditions that trigger actions (like sending you an email or changing updating a page in Notion).<p>Our long-term vision is that canvases become more interactive and interconnected, so that you end up building mini applications without it feeling like you're using a low-code app builder.</p>
]]></description><pubDate>Mon, 05 Feb 2024 19:56:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=39266220</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=39266220</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39266220</guid></item><item><title><![CDATA[Show HN: I made an Apple Vision Pro review roundup with AI using Jumprun]]></title><description><![CDATA[
<p>We recently launched Jumprun as a private beta, and made this public canvas showing a roundup of Apple Vision Pro reviews. Let me know if you're interested in trying out Jumprun to build your own canvas and I'll get an invite to you. We're keeping it as a closed beta for the moment to control costs.</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=39265960">https://news.ycombinator.com/item?id=39265960</a></p>
<p>Points: 4</p>
<p># Comments: 5</p>
]]></description><pubDate>Mon, 05 Feb 2024 19:38:34 +0000</pubDate><link>https://jumprun.ai/share/canvas/01HNXB2K3GM7KPRP45Y2CVVJSC</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=39265960</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=39265960</guid></item><item><title><![CDATA[New comment by benlm in "FAA investigating controversial crash video"]]></title><description><![CDATA[
<p>I've been a skydiver for many years and as far as I know the FAA _does_ require the pilot to wear a bail out rig. Most skydiving pilots I know do wear them, but I have come across some pilots not wearing them in flight (even though their bail out rig was in the plane next to them).</p>
]]></description><pubDate>Wed, 29 Dec 2021 22:44:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=29731124</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=29731124</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=29731124</guid></item><item><title><![CDATA[New comment by benlm in "Finland has slashed homelessness; the rest of Europe is failing"]]></title><description><![CDATA[
<p><a href="https://www.sbs.com.au/topics/voices/culture/article/2017/07/04/us-and-them-what-homelessness-looks-around-world" rel="nofollow">https://www.sbs.com.au/topics/voices/culture/article/2017/07...</a><p>"Homelessness then, in Australia, is more than lacking a roof over your head, it is also the absence of those features associated with “home”: permanence, security, and the freedom to come and go."<p>"If the world were to accept Australia’s definition and include everyone with inadequate shelter, the number would exceed 1.6 billion – roughly 20 percent of the population. Also excluded from official figures are the world’s 65 million displaced refugees in temporary accommodation."</p>
]]></description><pubDate>Mon, 25 Jan 2021 22:00:45 +0000</pubDate><link>https://news.ycombinator.com/item?id=25909523</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=25909523</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=25909523</guid></item><item><title><![CDATA[New comment by benlm in "Coil Whine on New XPS 13 (9360) Is Widespread"]]></title><description><![CDATA[
<p>Received my new XPS 13 yesterday. Coil whine is very loud under high graphics load (such as watching a 4K video) to the point where you can hear clear clicking and screeching several feet away. Disabling Intel TurboBoost in the BIOS made it much quieter, but you can still clearly hear it in a quiet room. It's unfortunate, because apart from this huge issue I love the laptop.</p>
]]></description><pubDate>Thu, 19 Jan 2017 02:32:08 +0000</pubDate><link>https://news.ycombinator.com/item?id=13432473</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=13432473</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=13432473</guid></item><item><title><![CDATA[Coil Whine on New XPS 13 (9360) Is Widespread]]></title><description><![CDATA[
<p>Article URL: <a href="http://en.community.dell.com/techcenter/os-applications/f/4613/t/19992655">http://en.community.dell.com/techcenter/os-applications/f/4613/t/19992655</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=13432426">https://news.ycombinator.com/item?id=13432426</a></p>
<p>Points: 3</p>
<p># Comments: 2</p>
]]></description><pubDate>Thu, 19 Jan 2017 02:21:28 +0000</pubDate><link>http://en.community.dell.com/techcenter/os-applications/f/4613/t/19992655</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=13432426</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=13432426</guid></item><item><title><![CDATA[New comment by benlm in "Ask HN: Idea Sunday"]]></title><description><![CDATA[
<p><a href="http://www.codetriage.com/" rel="nofollow">http://www.codetriage.com/</a></p>
]]></description><pubDate>Sun, 06 Apr 2014 18:54:42 +0000</pubDate><link>https://news.ycombinator.com/item?id=7542129</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=7542129</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=7542129</guid></item><item><title><![CDATA[New comment by benlm in "Ask HN: Idea Sunday"]]></title><description><![CDATA[
<p>examine.com meets google scholar<p>Place for all scientific studies to be elucidated with plain english explanations, analysis, information on shortcomings, links to related studies, open questions, links to further information. Would be community created content, but needs some sort of community voting/moderation system to surface the best content and empower the most knowledgable/trusted contributors.<p>Ideally this would be where you go when you see sensationalist headlines in popular science magazines -- here you can find out what the study actually means.<p>Anything like this out there?</p>
]]></description><pubDate>Sun, 06 Apr 2014 18:04:05 +0000</pubDate><link>https://news.ycombinator.com/item?id=7541821</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=7541821</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=7541821</guid></item><item><title><![CDATA[New comment by benlm in "Amazon Fire TV"]]></title><description><![CDATA[
<p>It's not actually clear whether Amazon Instant Video will stream at 1080p for any/all HD material, or if the device just supports 1080p for other services like Netflix. Does anyone know?</p>
]]></description><pubDate>Wed, 02 Apr 2014 19:28:30 +0000</pubDate><link>https://news.ycombinator.com/item?id=7518296</link><dc:creator>benlm</dc:creator><comments>https://news.ycombinator.com/item?id=7518296</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=7518296</guid></item></channel></rss>