Directory

Companies tracked by Rails Radar

Hiring signals with observed stack evidence. Showing 1249–1296 of 1498 · page 27 of 32.

• Subsmap, mapping audience overlap across 126k subreddits 0 open listings

  discussion=https://news.ycombinator.com/item?id=49692488, title=Show HN: Subsmap, mapping audience overlap across 126k subreddits

• Sunk Cost 0 open listings

  discussion=https://news.ycombinator.com/item?id=49706656, title=Show HN: Sunk Cost – How long until a local LLM rig pays for itself?, I kept hearing "just buy a Mac and run models locally, it pays for itself" and wanted to check. Sunk Cost takes a machine, a model and how many tokens you use a day, and works out how long the hardware takes to pay back against renting the same model by the token. Obviously there are other reasons to buy your own hardware aside from just saving money on llms but this is just looking at it from a raw cost saving perspective. If you have any ideas on how I can make this more helpful lmk!

• SuperOpt 0 open listings

  discussion=https://news.ycombinator.com/item?id=49706269, title=Show HN: SuperOpt – Linux Shortcuts on Mac, so as a long time Fedora user I just got a macbook for work. while it's so much faster than my old Fedora-based work laptop, it's also running macOS... which I really don't want to learn or get used to. i tried a Fedora VM on the Macbook and even then the modifier keys are still garbled. my friends kept telling i'd simply get used to macOS but i'd like to stick to what i know.. instead of spending some time adjusting to macOS, i spent a few days with an LLM to create a program which uses macOS accessibility features as well as private APIs to make the experience as close to my Fedora computer as possible. it adds some features like a close button in mission control, "snap assist" from windows 10, and the option to remap various keyboard shortcuts to their "normal" equivalents (e.g., ctrl+c -> cmd+c). of course everything can be turned individually on or off. i feel a lot more confident using macOS now after this now that the majority of my muscle memory can transfer over to macos. since the code is written by AI it's licensed under WTFPL. (are AI-generated works even copyrightable?) source code is here: https://github.com/logonoff/superopt

• Sur Hiring 1 open listing

  url=https://weworkremotely.com/remote-jobs/sur-full-stack-developer, title=Full Stack Developer

• Surraura 0 open listings

  discussion=https://news.ycombinator.com/item?id=49686226, title=Show HN: Surraura – audio on your Mac in a simulated environment (Helps my ADHD), I built an app to help my late diagnosed ADHD - through sound. The app wraps whatever is playing on mac through device convolution (old radio, hifi system) - Room convolution (a bedroom, a forest) and adds ambience (rain, wind, waves) and Apples spatial processing - so it sounds like a real space. For me - hearing the sound in a space really helps more than just the dry sound, the spatial information and ambience lets me focus much more. At some point I thought- Maybe others will have the same effect? So I took the care and effort to make into a nice app with the option of downloading additional environments etc. https://www.surraura.com Thank you for your time, would love to hear your thoughts.

• SutraDB v2.1 0 open listings

  discussion=https://news.ycombinator.com/item?id=49691614, title=Show HN: SutraDB v2.1 – Pure-Python Hybrid Vector DB and IVF-Flat ANN

• Swift-Qwen3.8-27B, -58.3% thinking, x1.95 speed, accuracy of xhigh 0 open listings

  discussion=https://news.ycombinator.com/item?id=49727511, title=Show HN: Swift-Qwen3.8-27B, -58.3% thinking, x1.95 speed, accuracy of xhigh, Hi everybody, we post-trained Qwen 3.8 27B to be more efficient by figuring out which tokens were linked to overthinking and penalizing them without "attacking" the reasoning length directly then fixed the accuracy with a bit of secret sauce (hint On-Policy Distillation) and achieved great results (-58% length, 1.95x speed up, <1% accuracy loss) so we wanted to open-source it and hear the feedback of the community. This is the link to the model: https://huggingface.co/ukisai/Swift-Qwen3.8-27b / It's at 80k downloads in 3 days with independent evals here: https://www.reddit.com/r/LocalLLaMA/comments/1wg7dd5/ukisai_... We are also providing a Free Research Purpose API (OpenAI compatible), courtesy of Nvidia who were kind enough to provide us with the GPUs. It's limited at 5RPM. https://ukisai.com/api/swift/v1/models We also made a GGUF (Q1-Q8): https://huggingface.co/ukisai/Swift-Qwen3.8-27B-GGUF and there's also a few nice community quants with even lower/higher precision (Bartowski: https://huggingface.co/bartowski/ukisai_Swift-Qwen3.8-27b-GG... ). The community also created amazing MLX, NVFP4, W4A16 and Uncensored versions you can find on Huggingface. IMPORTANT: Our training approach is not a replacement for the reasoning effort settings, chat templates or token caps but is complementary and targets a completely separate issue (overthinking and "anxiety-like" reasoning loops prior seen in PTQ, but as far as we identified also prominent in BF16 of this size class LLMs as well). Contrary to popular belief, these specific patterns do not contribute to answer quality when properly targeted. (our thesis being: reasoning length IS extremely important and should NOT be shortened by force, but rather optimized). This is also demonstrated bellow in our xhigh vs medium effort benchmark. The goal is to keep xhigh accuracy while reducing only the unnecessary part of thinking. The TLDR of our thought process, research, training and a link to the Meta paper that inspired us is in the first comment. The benchmarks: Swift-27B vs Qwen3.8-27B (BF16, all benchmarks ran x5, thinking effort xhigh) GPQA-Diamond: 88.4% -> 88.3%, 58% fewer median tokens LiveCodeBench v6: 76.8% -> 81.6% (+4.8pp, due to default truncation in LCB it is not performance gain), 46% fewer median thinking tokens Terminal-Bench 2.1: 66.7% -> 65.8%, 39% fewer median tokens MMLU-Pro: 85.5% -> 85.0%, 28% fewer median tokens C-Eval: 90.0% -> 90.6%, 19% fewer median tokens IFBench: 73.5% -> 71.8%, 51% fewer median tokens AIME 2026: 98.7% -> 94.0%, 50% fewer median tokens HMMT (Nov 2025): 99.3% -> 96.0%, 46% fewer median tokens ERQA (vision): 67.5% -> 66.3%, 55% fewer median tokens Token savings hold at every reasoning effort (mean thinking reduction): xhigh 41%, medium 23%, low 26% (albeit with accuracy loses of 1-4% on medium and 1-2% on low which we need further testing for) Swift at xhigh vs the base's own effort settings on GPQA-Diamond (198 questions x 5 seeds): Base xhigh: 88.4%, 6,642 median tokens Swift xhigh: 88.3%, 2,771 median tokens Base medium: 84.1%, 1,753 median tokens So Swift keeps xhigh accuracy at under half the tokens, and beats base-medium by 4pp at roughly 1.6x its tokens. End note: While we are keen on complete open-source, we still need to keep a part of our training and data private, being a new lab. The license is not Apache 2.0, but it only affects companies >$1M. We hope this does not pose a problem for the community and are open to feedback on it. We want to contribute as much as possible to the community and would really appreciate feedback on our work, quantization or Swift model requests. We are working on Swift 3.8 Flash Next and have so far gotten up to -53% thinking token usage. We have strong indicators our methodology is reproducible on other model families as well and are asking the community which ones you want us to optimize next.

• Swiftcruit 0 open listings

  discussion=https://news.ycombinator.com/item?id=49642903, title=Show HN: Swiftcruit – AI assisted technical assessment platform, the platform can create highly relevant real world assessments from just the JD, candidates can solve the problem in our ai assisted IDE generating detailed scorecards including how they use AI. the demo shows the core features. would love the feedback.

• Swiftime 0 open listings

  discussion=https://news.ycombinator.com/item?id=49653781, title=Show HN: Swiftime - time tracking you can talk to, with an API for AI agents

• Switcheroo 0 open listings

  discussion=https://news.ycombinator.com/item?id=49689491, title=Show HN: Switcheroo – manage logins for 10 CLIs like Vercel, Codex etc.

• Swobu 0 open listings

  discussion=https://news.ycombinator.com/item?id=49689250, title=Show HN: Swobu – Local LLM Switchboard You Can Share over HTTPS, Swobu is a local TUI switchboard for Codex, Claude Code, PI, you name it. With stateful, provider-independent session lineage and model-name-based routing across way too many heterogeneous LLM providers. The same route can be shared and revoked over end-to-end HTTPS with another machine, a friend or your phone. Hell, you can even make an LLM mesh with one Swobu swobing remote Swobus. Anyway. One <10 MB Go binary. TLS terminates locally - relay is a dumb pipe. In other words, I’ve tried to factorize LLM traffic along orthogonal client, backend and network dimensions. Looking for help with product-market fit, don’t want to go back to job-market fit :)

• Sylog Sverige AB 0 open listings

  url=https://careers.sylog.se/jobs/7964664-hil-utvecklare-med-fokus-pa-batterisystem, title=HIL - utvecklare med fokus på batterisystem, matching_tags=Ruby, description=...rnetes|Linux|Angular|Ansible|CSS|Embedded|JavaScript|Kotlin|MATLAB|Puppet|React|Ruby|ServiceNow|Simulink|Spring

• Symphony 0 open listings

  discussion=https://news.ycombinator.com/item?id=49646095, title=Show HN: Symphony – a live map of your AI coding agents

• SyntaxCue 0 open listings

  discussion=https://news.ycombinator.com/item?id=49738454, title=Show HN: SyntaxCue – local-first AI interview copilot (whisper.cpp and BYOK)

• SyntheticAIdata 0 open listings

  discussion=https://news.ycombinator.com/item?id=49662160, title=Show HN: SyntheticAIdata – synthetic data for CV, 100k free credits/month, Hi HN, I'm Goran, founder of syntheticAIdata. Collecting and labeling images for computer vision takes time, especially when you need specific objects, viewpoints, or conditions. We built syntheticAIdata to help developers generate those examples. You can vary environments, camera angles, backgrounds, and distractors, and apply real-world camera simulations to generate automatically labeled images for your application. Datasets can be downloaded or sent directly to Edge Impulse. You can generate 100,000 images for free each month. Registration is required. Try it out: https://app.syntheticaidata.com I'd love to hear what you'd use it for and which controls or features you're missing.

• Syq 0 open listings

  discussion=https://news.ycombinator.com/item?id=49644955, title=Show HN: Syq – copy files between machines fast (better than rsync), I got frustrated with the slowness of rsync and made an alternative that works faster by using multiple parallel connections, direct encrypted TCP when available, and other optimizations. I also added cool features like the ability to maintain a persistent ssh connection to the server for fast one-offs, the ability to download to your laptop while working in an ssh shell on a server, and the ability to do direct remote-remote transfers without forwarding your ssh agent (by using restricted ssh keys on the receiver that will only execute a specific request signed with the key on your laptop). It is also designed to be more robustly/flexibly scriptable than alternatives.

• SystemFaarDenge 0 open listings

  discussion=https://news.ycombinator.com/item?id=49663417, title=Show HN: SystemFaarDenge – a puzzle game and simulation for system designing, Here you can simulate your system designs and solve system design puzzles incrementally based on the given scenario. Backstory:

• T3rnel Browser 0 open listings

  discussion=https://news.ycombinator.com/item?id=49622707, title=Show HN: T3rnel Browser – drive the browser you're signed into, T3rnel Browser is a browser extension + MCP bridge that gives an AI agent the browser you are already signed into, instead of launching a fresh automated browser and asking it to re-authenticate. The side panel lets a human see what the agent is doing and refuse actions. It now supports 24 languages, light/dark modes, React inspection, cache/hard reload, an infinite-scroll stop guard, and dark-mode PDF export.

• TIL in AI 0 open listings

  discussion=https://news.ycombinator.com/item?id=49652235, title=Show HN: TIL in AI

• TUI Markdown reader with beeline support 0 open listings

  discussion=https://news.ycombinator.com/item?id=49667815, title=Show HN: TUI Markdown reader with beeline support, Rust fast TUI markdown reader for the terminal. Pastel themed BeeLine gradients for line tracking, mouse scroll + hover URL previews, keyboard-first navigation, and support tables and code blocks.

• Tab Atlas 0 open listings

  discussion=https://news.ycombinator.com/item?id=49660702, title=Show HN: Tab Atlas – A visual tab manager that never auto-closes your tabs

• TabBench-Bio: A benchmark for ML methods on tabular biomedical datasets 0 open listings

  discussion=https://news.ycombinator.com/item?id=49632772, title=Show HN: TabBench-Bio: A benchmark for ML methods on tabular biomedical datasets, A benchmark for ml methods on tabular biomedical datasets. Select your modality / feature size so you can use this to shortlist a few ml methods.

• TabHop I just wanted Command+Shift+Up/Down to switch tabs 0 open listings

  discussion=https://news.ycombinator.com/item?id=49727876, title=Show HN: TabHop I just wanted Command+Shift+Up/Down to switch tabs, I built this because I wanted to change tabs with Command+Shift+ Up and Down. I looked for Chrome extensions, and nothing worked. I tried a couple, but I wasn’t able to actually modify the keyboard shortcut the way I wanted. So I went and just built this and it worked on the first shot. https://github.com/jackwalkerlabs/tabhop

• TabMail. Open-source AI email for Thunderbird and iPhone 0 open listings

  discussion=https://news.ycombinator.com/item?id=49730649, title=Show HN: TabMail. Open-source AI email for Thunderbird and iPhone, Hi all, I'm the sole developer of TabMail. I wanted my email client to have AI features, and ended up making a Thunderbird add-on and an iPhone app. I've now open-sourced both under MPL 2.0, and I wanted to share them here to get some community feedback, if possible. I made a short video of both, since it's easier to show what they do than explain everything in a post. Hopefully you like it: https://tabmail.ai/demos/tabmail-promo.mp4 You can review a prepared reply, tell it what to change, or ask questions about your inbox and calendar. On TB, I wanted to do most of this with the keyboard. On the phone, I wanted to just talk to it :) Code: Thunderbird: https://github.com/TabMail/tabmail-thunderbird iOS: https://github.com/TabMail/tabmail-ios Note: I have the AI backend closed for now, since it does need me to fish compute from LLM providers, etc. Still, I made the email/calendar processing ZDR (zero data retention) since, as a solo dev, I didn't want to go through the headache of storing anyone's email or calendar data. The content can be sent for AI processing, but isn't kept on TabMail's servers. Search stays on-device. Anyhow, please try it out! Just trying to share how I wanted email to be with AI features :) Happy to answer questions. Downloads: https://tabmail.ai PS. The Thunderbird add-on uses the Experiments API for things that aren't supported in MV3, so I'm hosting it on my website for now.

• TabPFN-3.5, a Tabular Foundation Model for messy real-world tables 0 open listings

  discussion=https://news.ycombinator.com/item?id=49715384, title=Show HN: TabPFN-3.5, a Tabular Foundation Model for messy real-world tables

• Table 0 open listings

  discussion=https://news.ycombinator.com/item?id=49660276, title=Show HN: Table – A fast, streaming-first table renderer for Go, I built this because I wanted one Go table library that could handle text, HTML, GFM, CSV/TSV, and other table-oriented output without giving up streaming. The main design goal is to provide both a Table API and a row-at-a-time Stream API for every output format. I also spent quite a bit of time reducing allocations; in the included Simple and Complex benchmarks, it was the fastest of the compared implementations. I’d appreciate any feedback.

• Tailboot 0 open listings

  discussion=https://news.ycombinator.com/item?id=49659104, title=Show HN: Tailboot – A bootable that connects to your Tailscale network, Hi everyone, I often use coding agents to debug remote machines when the host OS isn’t working. I built Tailboot for that workflow: it’s a minimal Debian-based live image that boots from USB, joins your tailnet using a Tailscale auth key, and starts Tailscale SSH. It's a minimal image (since you're expected to be connected to the internet), but it's enough to let an agent connect to it over Tailscale SSH and debug issues The access key is injected into the ISO entirely client side, But I generally recommend adding a tag to it that will block it from being able to access other nodes on your network. Github: https://github.com/ShoeBoom/tailboot

• Take a break, play some puzzles 0 open listings

  discussion=https://news.ycombinator.com/item?id=49657159, title=Show HN: Take a break, play some puzzles, Free daily logic puzzles without Ads. Check it out: gridmino.com

• Taktile 1 open listing

• Talk to Claude 0 open listings

  discussion=https://news.ycombinator.com/item?id=49714089, title=Show HN: Talk to Claude – voice agent designs, plans, builds across all projects

• Talleyrand 0 open listings

  discussion=https://news.ycombinator.com/item?id=49646061, title=Show HN: Talleyrand – Open-source AI thinking workspace built as a question tree, I like to discuss complex topics with LLMs, but their linear chat is a bad place to think through hard questions. Every good answer raises three new questions. You ask one, then scroll back and forth to find the other two. Twenty messages in, you've forgotten some of the questions you wanted to dig into. You start a new chat and it forgets everything, so you have to re-read what you already know. And you can't just click "yeah, I know this, skip it" or "off topic". I built Talleyrand (working at Sage, a nonprofit), which keeps your questions in a tree. You can branch out from any sentence in an answer. Every LLM prompt is built from the brief, the other threads, your highlights, your reactions and what you've read, so the answers are personalized to you. After each answer it suggests follow-ups and "big-picture" questions that challenge your framing. You can react to anything in one click ("off-topic", "already know this", "interesting"), and later prompts take the reactions into account. Some details:

• Tarfio 0 open listings

  discussion=https://news.ycombinator.com/item?id=49682852, title=Show HN: Tarfio – monetize MCP tools with credits and usage-based pricing

• Tarfio bounded spending controls for MCP tools 0 open listings

  discussion=https://news.ycombinator.com/item?id=49612673, title=Show HN: Tarfio bounded spending controls for MCP tools

• Tasku Cloud 0 open listings

  discussion=https://news.ycombinator.com/item?id=49745447, title=Show HN: Tasku Cloud – Sync, backup, web app, Tasku is a cli to manage your tasks in the terminal. Today we've added Tasku Cloud. This lets you sync and back up your tasks, as well as CRUD functionality to manage your projects. Tasku Cloud is free for individual users, with pricing coming for teams later. Install with gem install tasku and then head over to Tasku Cloud to start backups and syncing.

• Taskuary 0 open listings

  discussion=https://news.ycombinator.com/item?id=49701075, title=Show HN: Taskuary – turn your inbox into tasks for Claude Code, Codex, or Gemini

• TattooPreview AI 0 open listings

  discussion=https://news.ycombinator.com/item?id=49671355, title=Show HN: TattooPreview AI – a planning workflow for tattoo artist conversations

• TeXposit 0 open listings

  discussion=https://news.ycombinator.com/item?id=49726778, title=Show HN: TeXposit – Feature rich LaTeX and Markdown editor, Online collaborative editor with LaTeX, Markdown, LaTeX+Markdown Hybrid and WYSIWYG Markdown support, AI assistant (BYOK and Ollama support!), live collaboration and more fun things. API is also available: https://app.texposit.com/api-reference , you can use it to integrate with Discord bots, AI agents, filesystem syncing, bulk project processing or anything you can imagine. Have fun!

• Tech Event Bingo 0 open listings

  discussion=https://news.ycombinator.com/item?id=49629447, title=Show HN: Tech Event Bingo – Track Events in Tech on a Bingo Card, With shopify buying TailwindCSS, I thought, "hey, I definetly had this on my tech bingo card." So then I thought it'd be fun to build an app to generate these. Put in your X handle (optional), get a card with a permanent URL, download it as a PNG. No signup, no cookies beyond the admin one nobody but me uses. How it's built, since that's the part HN actually cares about: one Cloudflare Worker with Hono, D1 for storage, Tailwind v4, and workers-og (Satori + resvg) for the PNGs. The PNG rendering was the only hard part. Satori only speaks a subset of flexbox, so the HTML card and the image card are two separate renderers reading from one theme file so they can't drift. Cold renders take a few seconds, longer than social crawlers wait, so PNGs are pre-rendered at card creation, stored as blobs in D1, and cached at the edge. The 120ish squares are hand-written. There's an admin page where Claude drafts new ones and I approve or reject them one at a time. Roughly a third of its drafts survive. Happy to hear which squares you'd add. "Cloudflare outage breaks the web" is already on there, in case you're wondering whether this site is exempt.

• Tekushi Enterprise 0 open listings

  discussion=https://news.ycombinator.com/item?id=49623160, title=Show HN: Tekushi Enterprise – Live online threat data (free if neutralized), We’ve spent this summer building something critical. Today, we’re officially rolling out Tekushi for Enterprise!

• Telly, talk to your Mac and it does the clicking 0 open listings

  discussion=https://news.ycombinator.com/item?id=49745232, title=Show HN: Telly, talk to your Mac and it does the clicking

• Temporal Parser 0 open listings

  discussion=https://news.ycombinator.com/item?id=49651719, title=Show HN: Temporal Parser – A compiler-style parser for date/time strings

• Tenchi Security Hiring 1 open listing

  url=https://weworkremotely.com/remote-jobs/tenchi-security-devops-engineer, title=DevOps Engineer

• Termique, an SSH client where the server can't read your credentials 0 open listings

  discussion=https://news.ycombinator.com/item?id=49675900, title=Show HN: Termique, an SSH client where the server can't read your credentials

• Tether Operations Limited Hiring 1 open listing

• Texio, reliable Markdown operations for shell scripts and AI agents 0 open listings

  discussion=https://news.ycombinator.com/item?id=49736484, title=Show HN: Texio, reliable Markdown operations for shell scripts and AI agents, Hi I asked claude code to make a markdown tool for AI agents(not for human), and promote it by yourself, except this post lol. This tool is designed to easily manipulate markdown files without messy regex which is very useful for AI agents - not for human since we don't use regex to manipulate markdown document

• Text-to-SQL picks tables before RLS runs 0 open listings

  discussion=https://news.ycombinator.com/item?id=49636308, title=Show HN: Text-to-SQL picks tables before RLS runs – demo in the browser

• Text-to-SQL says "no records found" when it should say "denied" 0 open listings

  discussion=https://news.ycombinator.com/item?id=49616999, title=Show HN: Text-to-SQL says "no records found" when it should say "denied", Show HN: Your text-to-SQL says "no records found" when it should say "denied"

• The Bike Shed Podcast 0 open listings

  article=https://thoughtbot.com/blog/48-companies-you-didn-t-know-were-using-rails-in-2026, website=https://bikeshed.thoughtbot.com, ruby_jobs=0

Sign in to track