António Domingues

I have been called a "serious #DataScientist", truth is that I love playing with all kinds of data, first and foremost with #Biological #Data. Now doing my magic in a #Biotech / #Pharma start-up.

I used to be a wet lab scientist. I will always be a foodie fueled by Espresso.

#rstats first but whatever tool gets the job done including #bash, #python. #linux user.

António Domingues boosted:

A friend and I built mcsprit.de, a website to see real time gasoline prices in Germany.

A few years ago, we applied to the German cartel office, which provides an API for this data.

We intend to have some fun with time series prediction models.
It's wild how frequently petrol stations change their prices. But there's a piece of legislation under discussion to reduce the number of price changes per day 🤞

#civictech #solidjs #rust #ml #datascience

A screenshot of a website showing a search bar for location and zip code at the top, a map of Hamburg in the middle, with pins and prices for petrol stations, and a list of petrol stations below the map, showing the first two with their name, zip code, price and the time of the last price update. It’s 1,49€ at 21:20 for the first station and also 1,49€ at 21:00 for the second one. 
The petrol station names are: 

Raiffeisen Energie Nord GmbH at zip code 22952
and the second one is
Calpam Express 6-22h Tanken at zip code 22941
António Domingues boosted:
nixCraft 🐧nixCraft
2025-12-20

Microsoft Teams is experiencing issues, with thousands reporting problems sending messages, including delays 🤣

A meme where a young kid smirks at the camera while a house burns in. The burning house is labeled "Microsoft Teams" and the girl is labeled "30% AI-generated Code."
António Domingues boosted:
Bruno Rodrigues :rstats: :tux:brodriguesco@fosstodon.org
2025-12-20

Gotta hand it to Apple, the M-series macbooks are really amazing hardware: but they only reach perfection with #NixOS on them!

2025-12-20

@brodriguesco @kupac how easy was it to install Linux on it? I know it's NIXos, but close enough :)

António Domingues boosted:
2025-12-20
António Domingues boosted:
Henrik Bengtssonhenrikbengtsson
2025-12-20

Possible a new record for me;

CRAN submission of {futureverse} took just six (6) minutes from start to finish 👏

(it's a package with no reverse dependencies)

António Domingues boosted:
Barry Rowlingsongeospacedman
2025-12-20

@Lluis_Revilla it says "using generative AI to enhance the CRAN checks". Are you saying this is wrong?

António Domingues boosted:

Closing out my year with a journal editor shocker 🧵 Checking new manuscripts today I reviewed a paper attributing 2 papers to me I did not write. A daft thing for an author to do of course. But intrigued I web searched up one of the titles and that's when it got real weird...

António Domingues boosted:
Danielle Navarrodjnavarro@hachyderm.io
2025-12-20

yeah i hate this. pay your fucking staff properly, and incorporate that into the price of the service.

abc.net.au/news/2025-12-21/aut

António Domingues boosted:
2025-12-20

@xgranade beating motherfuckers over the head with this MP4 until the end of time

António Domingues boosted:
Klyx - 39C3 (☎️ 9784)klyx@troet.cafe
2025-12-20

@schenklklopfer ich hab letztens was ähnliches überlegt. Aber habe mich dann erstmal dagegen entschieden. Einerseits weil das offizielle Angebot für meine Produktiv-Anwendung eines Kunden sicherer ist und weil ich mich auch gewundert habe, ob es nicht eingeschränkt wird in irgendeiner Form. Aber offenbar haben das schon einige erfolgreich gemacht: github.com/legout/sbox-s3

António Domingues boosted:
2025-12-20
António Domingues boosted:
ᴅʀ. ᴍᴏᴡɪɴᴄᴋᴇʟ'ꜱDrmowinckels@fosstodon.org
2025-12-20

Day 20: Performance Testing with bench

Profile and benchmark your package functions.

Basic benchmarking:

results <- bench::mark(
old_approach = old_function(data),
new_approach = new_function(data),
check = FALSE, # Skip result equality check
iterations = 100
)
plot(results)

Pro Tip: Include benchmarks in your test suite to catch performance regressions.
Resources: bench.r-lib.org/

#rstats #RPackageAdvent2025

bench::mark(
  my_function(big_data),
  filter_gc = FALSE
)results <- bench::mark(
  old_approach = old_function(data),
  new_approach = new_function(data),
  check = FALSE,  # Skip result equality check
  iterations = 100
)
plot(results)
António Domingues boosted:
Schenkl | 🏳️‍🌈🦄schenklklopfer@chaos.social
2025-12-20

Statt 1TB #S3 Speicher bei #Hetzner für ~6€/Monat zu mieten vielleicht 1TB #StorageBox für ~4€/Monat bei Hetzner mieten, via SMB in einen Server mounten und per Garage oder so als S3 bereitstellen?

Ich meine wie schlimm kann das werden?

Ich fürchte zwar die Antwort ist "JA", aber hats schon mal wer versucht?

António Domingues boosted:
Bruno Rodrigues :rstats: :tux:brodriguesco@fosstodon.org
2025-12-20

Korppi , a local-only, open-source Markdown editor for asynchronous collaboration.

With Korppi, each document carries its full history, so you can import multiple versions of the same file and apply changes hunk by hunk, calmly and explicitly.

youtu.be/nDaliT9W_zU

António Domingues boosted:
2025-12-20

Ensembl is looking for a web developer with experience in #genomics (or #bioinformatics) and #rust.

If you’re interested, please pop in an application! Otherwise, please boost!

#jobs #getFediHired #science

embl.wd103.myworkdayjobs.com/e

Screenshot of a genome browser in the ensembl website
António Domingues boosted:
🦜 Expresso | Liberdade para pensarexpresso.pt@rss-parrot.net
2025-12-19

Caso MIT/Brown: o que se sabe sobre o suspeito português e a ligação entre os dois crimes

expresso.pt/sociedade/2025-12-19-caso-mit-brown-o-que-se-sabe-sobre-o-suspeito-portugues-e-a-ligacao-entre-os-dois-crimes-8c0d8ce3

Português suspeito de duplo homicídio nos EUA encontrado morto após ter cometido alegadamente os crimes

António Domingues boosted:
small robotssmolrobots
2025-12-19

On the ninth day of Christmas, my Bigbot gave to me...

Nine Ladybots Dancing...

A darkened disco, with an illuminated dancefloor of checkerboard red and green panels. Nine robots are dancing, and all of them are Followerbots – robots based on real followers of mine (from Twitter, for the most part) who won competitions and also are women. From left to right we have Hattiebot (an ovoid robot with a little ringed planet on her antenna); Muckypupbot (looks like a little dog splattered with mud which is leaving muddy footprints on the dancefloor); Bethbot (has a propellor flying in the air, with a mop of red hair and glasses); Internetmombot (FKA Internetdadbot, ovoid with long, brown hair); Rossybot (is wearing a sort of woolly parka hood); Hottoastbot (shaped like a desk lamp for reasons that escape me); Amandabot (has frizzy reddish hair and glasses); Doctorbethbot (ovoid with dark hair, a pith helmet, and a red bowtie); and Tutbot (spherical, holding a glass of gin and wearing a purple fedora, staring sternly over her glasses). Discobot (a disco ball with sparkly platform shoes) hangs from the ceiling above, and at the back Littlegreyboatbot (one of the Ladybots in the previous version, but now not) has crashed through the wall and Emmabot (spherical, freckles, a brown ponytail) is smiling up at it, while Snathebot (a Custombot based on the real LGB’s spouse) peers over the gunwales. Basically, there’s a lot going on.
2025-12-19

@pgreer this was not the vibe I got from the article. The author mentions clearly:

> Now obviously, the reason a large language model focused on coding is good at Javascript and C++ is precisely because it has been trained on billions of lines of code in those languages along with countless forum posts, StackOverflow debates, and so on.

So they know what is going on. What they posit, to their sadness, is at some points LLMs might be trained on so many prompts that a VOPL might emerge.

António Domingues boosted:
2025-12-19

@kupac @keyboardpipette (I am in that toot and feel seen!)

Seriously, been using foldable keyboards and Palm PDAs similarly.
(I love the 4 part W-shaped ThinkOutside. Later they also made Bluetooth and USB ones, and I am still using those to this day)

Client Info

Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst