@jh3yy — CSS scroll-driven animations for scroll indicators
CSS Tip! 📜
You can create scroll indicators with scroll-driven animations 📍 Reveal a shadow, mask the content, etc.
.indicator--top {
animation: reveal;
animation-timeline: scroll(nearest);
animation-range: 0 2rem;
}
@keyframes reveal { 0% { scale: 1 0; }}Smart, because this replaces the usual scroll-event JavaScript listener with pure CSS — no handler, no requestAnimationFrame, no state. animation-timeline: scroll(nearest) ties the animation progress directly to the scroll container, and animation-range lets you scope it to just the top 2rem. The browser does the math.
@LinusEkenstam — Groq running Llama 2 70B at ~500 tokens/sec
🤯 Holy smokes. I can't believe this.
Output from groq.com is so fast, ~500 T/s
It spits out things near real-time, running on custom hardware, and it's based on Llama 2 70b
Best thing, you can try it for yourself right now.
That's not a typo — 500 tokens per second is roughly 10x what you'd get from a typical GPU cloud provider running the same model. Groq's LPU (Language Processing Unit) is a purpose-built chip that sidesteps the memory bandwidth bottleneck that makes GPUs slow at autoregressive inference. The practical effect: streaming output that's basically instantaneous.
@steventey — Open-source script to programmatically index pages via Google API
Just discovered this amazing open-source script by @goenning that uses Google's Indexing API to programmatically index all your pages → https://t.co/bInoTSvgRo
Took me ~10 min to run the script and index a bunch of pages that were previously unindexed for @dubdotco 🤯
The Google Indexing API was technically only meant for job postings and livestreams, but people have been using it for general pages for years and it works. Running a script beats waiting weeks for Googlebot to crawl your sitemap organically — especially for pages deep in the hierarchy that rarely get linked to.
@LangChain — RAG From Scratch: Hypothetical Document Embeddings (HyDE)
RAG From Scratch: Query Translation (HyDE)
Our RAG From Scratch video series walks through impt RAG concepts in short / focused videos w/ code. This is our final video on Query Translation, focused on Hypothetical Document Embeddings (HyDE) from Gao et al.
Interesting because HyDE flips the usual RAG lookup: instead of embedding the raw user query and searching for similar docs, you ask the LLM to generate a hypothetical document that would answer the query, then embed that. The intuition is that hypothetical documents are closer in embedding space to real documents than raw queries are. Makes the retrieval step more robust for short or ambiguous questions.
@buninux — Responsive design shortcuts in Figma
cheating in responsive design with Figma
The word "cheating" is doing a lot of work here, but the technique is legit — using Figma's constraints and auto-layout to approximate responsive behavior without manually building every breakpoint. Honestly useful as a design-to-dev handoff shortcut when you just need to show intent fast.
@jasonleowsg — diskprices.com: $5k/month from a simple comparison page
What's stopping you from making your SaaS like this?
Fun fact: This site (diskprices.com) makes $5k/m
diskprices.com is one of those sites that's been around forever — it scrapes disk prices from Amazon and displays them in a sortable table. No auth, no SaaS, no onboarding flow. Makes sense as a reminder that "boring" tools with a very specific audience and no competition can quietly earn well without any of the startup overhead.
@rithulkamesh — Pomoflow: open-source collaborative pomodoro timer
I built pomoflow.io with @qoobes, an open-source, distraction-free collaborative pomodoro timer using @shadcn's ui and @Firebase, deployed on @vercel and @flydotio!
read the blog @ pomoflow.io/blogs/pomoflow-v1
Worth saving mostly as a reference for the stack: shadcn/ui + Firebase + Vercel + Fly.io for a real-time collaborative tool. The combination is practical and the blog post actually documents the build, which is rarer than it should be.
