🦞🌯 Lobster Roll

Stories by Beefin

Reverse Video Search (blog.mixpeek.com)
Show HN: Agent Multiplexer – manage Claude Code via tmux (github.com)
Applying Pandas' Group_by on Videos (mixpeek.com)
Show HN: FFmpeg.wasm In-Browser Video Conversion (ethan.dev)
Show HN: Simple CLI tools for multimodal data (github.com)
Al Jazeera Banned in the West Bank (ynetnews.com)
Show HN: A CLI tool for interacting with Linear (github.com)
Query Preprocessing: How we handle 500MB video files as search queries (mixpeek.com)
Why are embeddings so cheap? (tensoreconomics.com)
Show HN: IAB Taxonomy Mapper – Convert Taxonomies Using Ollama (github.com)
I built a free and open source tool to make working with the IAB content taxonomies less painful.<p>Converts IAB Content Taxonomy 2.0 → 3.0 automatically<p>Runs fully local, no data leaves your machine<p>Multiple methods supported (BM25, TF-IDF, embeddings, LLM re-ranking)<p>Simple demo UI + CLI ava...
Show HN: Intentflow – YAML-based UX engine with flags, tooltips, and LLM routing (github.com)
We built Intentflow <a href="https:&#x2F;&#x2F;github.com&#x2F;mixpeek&#x2F;intentflow">https:&#x2F;&#x2F;github.com&#x2F;mixpeek&#x2F;intentflow</a> to make UX flows declarative and adaptive.<p>Most product teams hardcode banners, modals, and onboarding steps. Intentflow lets you define those flows...
DIY Media Asset Management System (blog.mixpeek.com)
Show HN: Semantic Image Rabbit Hole (shimmer.so)
We run 20M models in parallel on Ray (mixpeek.com)
Show HN: Amux – A tmux-based multiplexer for running parallel Claude Code agents (amux.io)
Hi HN, I&#x27;m Ethan. I built amux because I was spending more time switching between terminal tabs than actually building things.<p>amux is an open-source agent multiplexer that lets you run, monitor, and control multiple headless Claude Code sessions from a single dashboard — in your browser, on ...
Show HN: CAD parts builder with WASM and WebGL (modo.is)
Show HN: Describe hardware, get a buildable prototype (modo.is)
I’ve built hardware projects for years, and the gap between “I want X” and “parts ordered, files ready to print” always took weeks. Modo compresses that. Describe what you want (e.g. “a portable CO2 monitor with a display and buzzer”) and it generates:<p>- A bill of materials with real components (M...
Show HN: Learn Vision Encoders (mixpeek.com)
Show HN: Multimodal Search over the National Gallery of Art (mxp.co)
We indexed 120K images from the National Gallery of Art for visual search. Text queries, image uploads, and &quot;find similar&quot; all in one retriever, fused with RRF.<p>Demo: <a href="https:&#x2F;&#x2F;mxp.co&#x2F;r&#x2F;nga" rel="nofollow">https:&#x2F;&#x2F;mxp.co&#x2F;r&#x2F;nga</a><p>Stack: S...
Show HN: Multimodal Benchmarks (github.com)
I built a set of open multimodal retrieval benchmarks because existing IR evals are still mostly text-only and don’t capture real-world complexity.<p>This repo includes ground-truth datasets, queries, and relevance judgments for 3 hard domains:<p>• Financial documents (SEC filings with tables, chart...
Learning to Rank with Bandit (mixpeek.com)
Show HN: Show HN: Geotagged Photo Map (ethan.dev)
I built a minimal personal site that’s entirely static—no frameworks, just `index.html`, `styles.css`, and a bit of vanilla JS. It has a geotagged photo album that renders a Leaflet map and a lightbox gallery from a simple `images&#x2F;gallery.json`. I wrote small scripts to bulk ingest photos, read...
Multimodal Late Interaction Models (mixpeek.com)
Embeddings Alone Can't Tame Video: Why Segmentation Is Key (mixpeek.com)
Awesome Multimodal (github.com)
Awesome Multimodal Search (github.com)
The Death of the Cookie (mixpeek.com)
MongoDB purchases Voyage AI for $220M (seekingalpha.com)
Ecstasy and Amnesia in the Gaza Strip (mosaicmagazine.com)
Show HN: Video Embeddings with Weaviate and Mixpeek (learn.mixpeek.com)