Antonio Monteiro

I am now on Github Sponsors. If you appreciate my work and want to support the projects that I work on publicly, you can now show your generosity by supporting me directly! github.com/sponsors/anmon…

how can you not wake up everyday amazed at what ML has become - AI is really real. the wildest dreams of a million nerds came true. i just cannot understand the naysayers.

@PThorpe92 some folks love Adobe PSD #L108" target="_blank" rel="nofollow noopener">github.com/gco/xee/blob/4…

What is the all-time worst file format? is there anything out there worse than PDF? (from a dev perspective, not just user)

writing parseff is mega fun this is a parser for a key/value (with utf8 support)

Introducing an official node.js SDK for FFF file search This is an official API for the fastest and the most accurate fs search working out of the box in node providing file search, grep, and multigrep support. This is all you need to be 18x faster than ripgrep using only node

I’m very happy to present my toy research project: Sotaku! It's a neural net that automatically discovered the rules of sudoku and learned to solve them, achieving a new state-of-the-art score of 98.9% on one of the hardest sudoku datasets, while being agnostic to the game, and beating all other sudoku-optimized neural net architectures* Read more for fun motivations, plus some extremely unconventional discoveries, e.g. reverse curriculum consistently beating curriculum (!), emergent reasoning-like capabilities, and the future of traditional programming

modern terminals are very capable

haha crazy - full stack ocaml & melange, plus islands & hydration. i didn't even think to ask it to open the browser and test it, but sent me a picture anyway. crazy days

We can't wait to welcome @davesnx to React Miami this April! David will be sharing his insights on what transformed him from someone who uses React, to someone who understands why it works—and how to apply those principles far beyond React itself.

we run 18 million EC2 instances per month. At our scale, we see very rare bugs very frequently. Last week, we received *half* an HTTP request. Not a HTTP 206, literally half a request. Content-Length was 2350 bytes. Body was actually 1200 bytes, and was truncated mid json doc.

By default, the most popular Rust crate for Postgresql (tokio_postgres) waits for *2 hours!!* before timing out a dead connection. All because of a bad decision from 1989 🧵👇️

Recall is the API for meeting recording. We capture millions of hours of meetings per week, and process over 3 TiB/sec of raw video at our peak load. Operating at this level of scale demands extreme reliability, so we have extensive testing of all parts of our real-time video processing pipeline. However, testing real-time video processing is excruciatingly slow, since these pipelines need to run in "real time". To test a 1 hour video capture, the test needs to run for 1 hour. We have hundreds of tests, and even with parallelism our suite was taking way too long and not fully utilizing our CPUs. Our solution: speed up time :)

We forked Chromium and bolted on a lock-free, zero-copy, low-latency shared memory ringbuffer written in Rust We needed to IPC 100+ MB/s of raw video, and Chromium's WebSocket implementation is dreadfully inefficient

At 7:00am PT every day, load on our systems spikes 300% in 60 seconds Recall is the API for meeting recording. Everyone starts their meetings at the top of the hour, meaning we handle Black Friday level traffic spikes 20x every day. At our scale, this involves launching hundreds of thousands of EC2 instances within seconds of each other. We process over 3 TiB/sec of raw video at our peak load We're hiring systems engineers. If this sounds interesting to you, DM me your GitHub!

Are there really only 3 cracked engineers in SF (or willing to move here)? Can't believe you didn't send me your GitHub after such a compelling offer 🫠

At @recallai, we ran 18M EC2 instances last month. We use Rust to process 3TB/second of raw video in real-time. If you’re a cracked engineer in SF, DM me your GitHub – if you interview with us, we’ll gift you a free bundle of K&R’s C, TLPI and TCP/IP Illustrated Vol 1 🔥

@davesnx damn nested record types look really good

This is what pragmatism and 2 years of hard work (with a bunch of contributors) have done Proud of you folks

Just integrated @RecallAI's desktop SDK into ToDesktop Builder. In 30 mins you can take your web app and ship it as a desktop app with: ✓ Meeting detection and recording out of the box ✓ Automatic transcription ✓ Speaker Identification 👇 Full tutorial and repo in thread

We ran 9 million EC2 instances last month. We're hiring engineers in SF.

Big. Update. @recallai already powers billions of minutes of meeting recording through our Meeting Bot API (used by 2000+ companies, including HubSpot, ClickUp, and Apollo). But bots aren’t the only way to capture conversations. Today, we’re introducing the Desktop Recording SDK. It lets developers embed recording directly into their apps with no bots required. That means: 1/ no bot accounts showing up in meetings 2/ recordings that “just work” across any desktop meeting app 3/ transcripts, recordings, speaker names, and metadata delivered through the same @recallai API p.s. build Granola-style features into your app