Timothée Poisot
@ctrlalttim.com
Almost certainly one of the ecologists of all time.
AI/ML, biodiversity monitoring, viral emergence, open science, methodological anarchism
he/they
🧪 https://epic-biodiversity.org/
📰 https://buttondown.email/ctrl-alt-tim
AI/ML, biodiversity monitoring, viral emergence, open science, methodological anarchism
he/they
🧪 https://epic-biodiversity.org/
📰 https://buttondown.email/ctrl-alt-tim
we're getting somewhere
October 29, 2025 at 5:24 PM
we're getting somewhere
Rough draft, obviously, but I think I like this approach to bivariate legends.
October 29, 2025 at 1:47 PM
Rough draft, obviously, but I think I like this approach to bivariate legends.
OK, that's enough dataviz for tonight.
October 29, 2025 at 1:56 AM
OK, that's enough dataviz for tonight.
I was out of the country for less than 3 days. My reimbursement claim is 24 pages long, for 7 items. By the time I was done, I had spent about 2 hours on it. 🧪
October 28, 2025 at 9:05 PM
I was out of the country for less than 3 days. My reimbursement claim is 24 pages long, for 7 items. By the time I was done, I had spent about 2 hours on it. 🧪
I surprisingly enjoyed writing my NSERC discovery application this time. Now I need to clean up my CCV. I will enjoy this a lot less.
October 24, 2025 at 5:44 PM
I surprisingly enjoyed writing my NSERC discovery application this time. Now I need to clean up my CCV. I will enjoy this a lot less.
A pretty good day for the bike ride in the way that this is a pretty nice rollercoaster, I guess
June 24, 2025 at 5:26 PM
A pretty good day for the bike ride in the way that this is a pretty nice rollercoaster, I guess
I am adding a lecture on boosting to my Data Science for Biodiversity class, so I decided to spend a few hours writing AdaBoost from scratch. This was fun!
June 24, 2025 at 3:48 PM
I am adding a lecture on boosting to my Data Science for Biodiversity class, so I decided to spend a few hours writing AdaBoost from scratch. This was fun!
When reviewer 2 uses your own papers to criticize your manuscript during anonymous peer review
June 17, 2025 at 11:47 PM
When reviewer 2 uses your own papers to criticize your manuscript during anonymous peer review
Men will literally nuke basalt instead of going to therapy
June 11, 2025 at 3:41 PM
Men will literally nuke basalt instead of going to therapy
Let's get this proposal finished - 10% editing skills, 90% Ethiopian coffee
June 9, 2025 at 2:19 PM
Let's get this proposal finished - 10% editing skills, 90% Ethiopian coffee
I might actually be that kind of doctor?
June 6, 2025 at 4:22 PM
I might actually be that kind of doctor?
So @datadryad.bsky.social has a new system to prevent AI scrapping (fine), but apparently it blocks all requests not made by a browser, so downloading data automatically with curl or wget is now broken? And it's this weeb-inspired mascot that is to blame? Very cool! Love open science! 🧪
April 18, 2025 at 9:31 PM
So @datadryad.bsky.social has a new system to prevent AI scrapping (fine), but apparently it blocks all requests not made by a browser, so downloading data automatically with curl or wget is now broken? And it's this weeb-inspired mascot that is to blame? Very cool! Love open science! 🧪
OK spotify calm down I don't need this attitude
April 11, 2025 at 8:26 PM
OK spotify calm down I don't need this attitude
when I have to use trig functions
April 2, 2025 at 11:40 AM
when I have to use trig functions
@typst.app is UNREAL. Fully bespoke library for slides in approx. 100 lines of code.
April 1, 2025 at 5:52 PM
@typst.app is UNREAL. Fully bespoke library for slides in approx. 100 lines of code.
We had a LOT of fun at the Lego robot workshop today.
March 15, 2025 at 11:19 PM
We had a LOT of fun at the Lego robot workshop today.
me too, repl, me too
February 28, 2025 at 6:26 PM
me too, repl, me too
I am now fully onboard with the idea of using the legend of a map to provide additional information about the data.
February 25, 2025 at 10:05 PM
I am now fully onboard with the idea of using the legend of a map to provide additional information about the data.
And if you are keen to know more, we're giving a workshop on it at the 2025 @csee-scee.bsky.social meeting in Sherbrooke! Come joint us as we walk through some cool case studies. And check out the preprint: ecoevorxiv.org/repository/v...
February 18, 2025 at 5:39 PM
And if you are keen to know more, we're giving a workshop on it at the 2025 @csee-scee.bsky.social meeting in Sherbrooke! Come joint us as we walk through some cool case studies. And check out the preprint: ecoevorxiv.org/repository/v...
But one of my favorite things that the package can do is explanations: I collected a lot of functions I wrote to teach data science into a small package to do simple SDMs, and it's sort of bare bones but very easy to use, so learners can experiment with it.
February 18, 2025 at 5:36 PM
But one of my favorite things that the package can do is explanations: I collected a lot of functions I wrote to teach data science into a small package to do simple SDMs, and it's sort of bare bones but very easy to use, so learners can experiment with it.
It also plays nice with docs.makie.org/dev/ - and most of the objects the package knows about can be plotted directly. This is nice. I love @makie.org , favorite plotting package ever, 12/10.
February 18, 2025 at 5:34 PM
It also plays nice with docs.makie.org/dev/ - and most of the objects the package knows about can be plotted directly. This is nice. I love @makie.org , favorite plotting package ever, 12/10.
Many packages? Yes! The core design philosophy is that things should be easy to maintain, easy to upgrade, and should not break. Although the toolkit is currently seven packages, there are only three hard dependencies. The rest is all methods overloading, and well-documented interfaces.
February 18, 2025 at 5:29 PM
Many packages? Yes! The core design philosophy is that things should be easy to maintain, easy to upgrade, and should not break. Although the toolkit is currently seven packages, there are only three hard dependencies. The rest is all methods overloading, and well-documented interfaces.
First, you can head over to the documentation website: poisotlab.github.io/SpeciesDistr...
It has tutorials, small how-to explainers, and a reference for the many packages that make up the full toolkit.
It has tutorials, small how-to explainers, and a reference for the many packages that make up the full toolkit.
February 18, 2025 at 5:28 PM
First, you can head over to the documentation website: poisotlab.github.io/SpeciesDistr...
It has tutorials, small how-to explainers, and a reference for the many packages that make up the full toolkit.
It has tutorials, small how-to explainers, and a reference for the many packages that make up the full toolkit.
guys I don't think I'm doing gradient descent well
February 13, 2025 at 6:18 PM
guys I don't think I'm doing gradient descent well