Jeff Martin
LightNews LightNews
jeff.cuchazinteractive.org
Jeff Martin
@jeff.cuchazinteractive.org
72 followers 77 following 210 posts
Making a better internet for myself and friends.

Software engineer working in academia, and on my own tools, in Rust, Kotlin, Mojo, Python, Javascript, and more.

Works on nextPYP: software for analyzing CryoET (and EM) data from your web browser.
Posts Replies Media Videos
Pinned
jeff.cuchazinteractive.org
Jeff Martin @jeff.cuchazinteractive.org · Jul 7
Sooo.... I finally got around to making a logo for my SmolWeb project. It's a peer-to-peer network for the web. Like, host a website using the computer and internet you already have. Then people visit your website using the web browser they already have. No clouds needed! Sort of, it's complicated.
A Smol globe drawn in a cartoon style with thick black strokes and a cute, if somewhat silly, face.
Reposted by Jeff Martin
ponderingpothos.bsky.social
Take a look under the hood of the state of the art in grid interpolation in Rust and Python!

Come for the compile-time loop unrolling, stay for the profile-guided optimization!

jlogan.dev/blog/2025/11...
Plot showing speedup relative to scipy ranging from 40 to 200x
November 11, 2025 at 3:04 AM
jeff.cuchazinteractive.org
InterpN: Fast Interpolation, using Rust

jlogan.dev/blog/2025/11...

Cool story about how to go fast with numerical computing. And why many Pythonic approaches, even with native extensions, can be pretty slow.

My first thought though: can we go faster in Mojo? 😅
2025-11-10 InterpN: Fast Interpolation - James Logan's Personal Site
jlogan.dev
November 11, 2025 at 4:41 AM
Reposted by Jeff Martin
davidaugust.bsky.social
Happy logical operator Halloween! 🎃
Meme showing six logical operators illustrated with jack-o’-lantern images. trick OR treat trick AND treat trick XOR treat trick NOR treat trick NAND treat trick XNOR treat
October 31, 2025 at 4:55 PM
Reposted by Jeff Martin
anildash.com
Really positive sign that open source projects are standing up to do the right thing. Kudos to the Python folks: www.theregister.com/2025/10/27/p...
Python Foundation rejects $1.5M grant with no-DEI strings
: Foundation says it won't compromise policy of inclusivity even if that cash would've really helped
www.theregister.com
October 28, 2025 at 3:17 PM
Reposted by Jeff Martin
sjorsscheres.bsky.social
We are looking for a computational postdoc to work with us on new optimisation algorithms to make #RELION even better. Join our bubbly team at the @mrclmb.bsky.social in Cambridge, UK. 🤗 RTs appreciated.

mrc.tal.net/vx/appcentre...
Quick Check Needed
mrc.tal.net
October 29, 2025 at 9:17 AM
jeff.cuchazinteractive.org
PSA for my friends who play games on Windows: Microsoft now screenshots your games and sends the images to the cloud for AI stuff, and who knows what else.

If this bothers you, you can turn it off (for now). Or you can switch to a different OS, like Linux or Mac.

www.techspot.com/news/109983-...
Microsoft's Gaming Copilot automatically captures screenshots, but you can turn it off
Microsoft appears to be capturing large numbers of screenshots through Gaming Copilot, extracting text from them, and using the data to further train its AI models. A...
www.techspot.com
October 26, 2025 at 2:54 PM
jeff.cuchazinteractive.org
Oof! There's nothing quite like spending a couple frustrating hours debugging an issue in my code only to trace it down to a library function sometimes destroying its inputs, and sometimes not!

Thanks, FFT libraries!
October 14, 2025 at 6:03 AM
Reposted by Jeff Martin
actacrystd.iucr.org
October issue is now available! journals.iucr.org/d/issues/202...

The cover highlights the article on categorizing prediction modes within low-pLDDT regions of AlphaFold2 structures from the Richardson lab @duke-university.bsky.social : doi.org/10.1107/S205...

Image credit: Chistopher Williams
October 2, 2025 at 1:58 PM
jeff.cuchazinteractive.org
Go home, Javascript, you're drunk!

[14, 9].sort() returns [14, 9]

Because arrays elements are converted to strings before comparison! 🤔
September 30, 2025 at 10:12 PM
jeff.cuchazinteractive.org
Welp. We've reached the part of the story where I get the Very Official Work Email that says: If the government shuts down, you should keep working. Because you'll still get paid. Probably.
September 29, 2025 at 5:21 PM
Reposted by Jeff Martin
mara.bsky.social
Happy 10 year anniversary to this tweet! Has memory safety in C++ been solved yet?

x.com/CppCon/statu...
A tweet from @CppCon from 22 september 2015: "Herb Sutter says we are close to solving memory safety in C++ without runtime overhead. #CppCon"
September 22, 2025 at 5:43 PM
Reposted by Jeff Martin
filippo.abyssdomain.expert
A "fun" issue my professional communities are dealing with recently is where to hold conferences, if non-US people can't safely enter the US, and foreigners in the US can't safely leave it.

There is no good answer, and the equilibrium is having fewer foreigners AND conferences in the US.
September 20, 2025 at 9:54 AM
Reposted by Jeff Martin
rude1.blacksky.team
the photo they take of you before going through TSA is optional
you can just walk away after they return your ID
September 18, 2025 at 5:59 PM
jeff.cuchazinteractive.org
My bsky feed is slowly filling up with science and CryoEM posts. I never really found much of the tech crew here (guess they stayed on Mastodon?), but it's nice seeing a bunch of scientists now. 🔬
September 18, 2025 at 12:49 PM
jeff.cuchazinteractive.org
Woo! My lab's Nature Protocols paper is out today.

In a nutshell, the protocol shows how to use our Cryo-ET software to solve high-res protein structures.

I don't think any of the other authors are here, but congrats to Hsuan-Fu, Ye, and Wendy for all of their work!

www.nature.com/articles/s41...
In situ structure determination of conformationally flexible targets using nextPYP - Nature Protocols
The procedure for nextPYP facilitates the end-to-end processing of cryoelectron tomography data for high-resolution structure determination and conformational variability analysis.
www.nature.com
August 19, 2025 at 12:57 PM
Reposted by Jeff Martin
adamwiggins.bsky.social
😍 Eng manager job working on storage APIs, workers, etc for Firefox. Relevant to improving local-first apps in the browser! www.mozilla.org/en-US/career...
Mozilla Careers — Engineering Manager, DOM (Web Standards) or SpiderMonkey (Javascript) — Open Positions
Mozilla is hiring a Engineering Manager, DOM (Web Standards) or SpiderMonkey (Javascript) in Remote US, Security, Security, Security, Strategy, Operations, Data & Ads,…
www.mozilla.org
August 4, 2025 at 8:02 AM
Reposted by Jeff Martin
kagi.com
Orion is coming to Linux early next year! 🙂
July 20, 2025 at 2:10 AM
Reposted by Jeff Martin
jobboardsearch.com
📢 Tailscale is #hiring a Software Engineer, Networking (Rust)!

💰 $181K - $240K
🌎 Worldwide

🔗 http://jbs.ink/7tEO5SjTpfDw

#jobalert #jobsearch #remotejob #remotework #wfh #backend #softwareengineer #design
📢 Tailscale is hiring a Software Engineer, Networking (Rust)!
July 16, 2025 at 7:31 AM
Reposted by Jeff Martin
alisterburt.bsky.social
@jeff.cuchazinteractive.org and I wrote a GPU kernel in Mojo for cryo-EM forward projection this weekend as part of the Modular Hack Weekend. On our benchmark volume we generate ~500k projections per second, a ~35x speedup over my PyTorch GPU implementation 🤩

code: github.com/alisterburt/...
June 30, 2025 at 5:05 AM
jeff.cuchazinteractive.org
Holy shit, the new Mario Kart World soundtrack goes hard! 🎷
June 24, 2025 at 11:26 PM
jeff.cuchazinteractive.org
Still one of the greatest jokes of all time:

No, Mr. Die, I expect you to bond.
A photo of a die bonded to a circuit board using many many separate little bonding wires
June 17, 2025 at 4:01 PM
jeff.cuchazinteractive.org
Wow, it's not every day you write some Python code, run into an error, search for a solution online, and the creator of Python says just ignore the error.

I feel a little vindicated here. 😅

github.com/python/typin...
"BinaryIO" has no attribute "readinto" · Issue #659 · python/typing
Per python/typeshed#2166 and @gvanrossum's comment there.
github.com
June 15, 2025 at 10:10 PM
Reposted by Jeff Martin
chezvoila.com
The promise is AI/LLM is that it will make things easier, but every time I read about how it actually works, it's just a different kind of difficult.
andrew.heiss.phd Andrew Heiss @andrew.heiss.phd · Jun 5
Updated my "AI, LLMs, and bullshit" mini-lecture post in my dataviz class with a new section on vibe coding datavizs25.classes.andrewheiss.com/resource/ai-...
Vibe coding and learning Remember that LLMs are (philosophical) bullshit machines. They don’t care if the answer if gives you is wrong or right. They don’t care if the code runs or not. All they care about is if the answer looks plausible. That said, in part due to the massive amount of publicly available code that’s inside their training data, LLMs can work okay with code in limited circumstances. Think of them as fancy documentation—if a programming language or package has good documentation with lots of good examples online, the stuff that LLMs generate about it will generally have good coverage and won’t be completely made up. They work well enough that a whole new approach to programming has emerged: vibe coding. When vibe coding, you essentially tell an LLM what you want it to make and then let it actually build the whole thing. Some code editors actually allow LLMs to create and edit files directly on your computer. Like, you can say Make a website that has a login page where users can log in and then create and edit blog posts. Create a theme for it that is mostly blue, but with some nice contrasting colors, and use a warm, friendly sans serif font. Add some photos from Flickr too. … and it will attempt to do that, creating all the necessary files and code. It feels neat and magical to code without coding. If you know what you’re doing, and you mistrust everything the LLM spits out, and hand-check everything it says, vibe coding can save you some time. HOWEVER it is (1) potentially dangerous, and (2) bad for learning. Danger! Because LLMs are designed to make plausible looking code, they can reference external packages that don’t exist. For instance, you might want to make a lollipop chart, and it’ll say something like Certainly! First, run remotes::install_github("someone/gglollipop") and then run library(gglollipop) to load it. BUT GUESS WHAT. There is no package named {gglollipop}. It’s not a thing. Malicious actors also vibe code and look for the names of packages that LLMs think should exist, then they go and create those packages themselves, but include malware in those fake packages. If you’re using an editor that allows LLMs to make changes to your computer, it can run all this code without you knowing what happens and bad things can happen. Additionally, vibe-coded code can run well but be full of security vulnerabilities and issues and inefficiencies and bugs. The code it generates is essentially an average of all other code online—it will have bugs and security issues. It’s also incredibly easy to accidentally expose private information to LLMs, like feeding them passwords, usernames, private data, and so on, especially if you’re mindlessly generating and running code. Always distrust and be suspicious about all code that LLMs generate. Bad for learning! Some people claim that vibe coding helps them learn how to code. This is most likely not the case—pure vibe coding likely hurts learning. Think of learning to code like learning a language. You could try to learn French by watching hours and hours of French movies, and you’d certainly pick stuff up, but you won’t learn nearly as well as you would by taking a dedicated French class. You need the foundation first. If you take a few years of French classes and then watch hours and hours of French movies, you’ll learn a lot. If you jump right into a pure media-based approach to learning, it’ll be detrimental to learning the language. Recent research explores this idea with LLMs: Students who substitute some of their learning activities with LLMs (e.g., by generating solutions to exercises) increase the volume of topics they can learn about but decrease their understanding of each topic. Students who complement their learning activities with LLMs (e.g., by asking for explanations) do not increase topic volume but do increase their understanding. We also observe that LLMs widen the gap between students with low and high prior knowledge. (Lehmann, Cornelius, and Sting 2025) Prior knowledge is essential if you want to do any code-related stuff with LLMs, and even then, LLMs are helpful when asking for explanations (e.g. “This code works, but I don’t know why! Can you explain what’s happening with geom_point() here?”), but far less helpful for learning than just generating plausible-looking answers. I can typically tell when you submit code generated by an LLM; it has a certain vibe/style to it and often gives you extraneous code that you don’t actually need to use. Like this—this comes right from ChatGPT: # Load the required libraries library(dplyr) # Read a file named cars.csv into the R session data <- read.csv("data/cars.csv") # Calculate the average value of cty by class average_cty_by_class <- data %>% group_by(class) %>% summarize(average_cty = mean(cty, na.rm = TRUE), .groups = "drop") # Show the results print(average_cty_by_class) The “tells” Here’s how I can tell that that ↑ code comes from an LLM: Comments at every stage read.csv() instead of read_csv(), The older %>% pipe instead of the newer |> pipe na.rm = TRUE inside mean() .groups = "drop" inside summarize() Using print() to show the created object In general, don’t ever just copy/paste from ChatGPT and hope it runs. If you do use these tools, use them to ask for explanations. Make sure you know what every line of code is doing. For example, if it spits out summarize(average_cty = mean(cty, na.rm = TRUE), .groups = "drop"), and you don’t know what those na.rm = TRUE or .groups = "drop" things mean, look at the help page for mean() and see what na.rm actually does so you can decide if you really need to use it or not, or ask the LLM why it included it. I will call out code that looks vibe-coded in your assignments. Please don’t do it.
June 5, 2025 at 3:08 PM
jeff.cuchazinteractive.org
Wow, C++ build systems are an absolute nightmare. 😱 I don't miss this at all. Rust's build tool Cargo has completely spoiled me.
May 23, 2025 at 2:10 PM
Reposted by Jeff Martin
signal.org
We're ready for Microsoft Recall and the automatic screenshots it takes of everything on your desktop.

Signal Desktop on Windows now includes support for a new "Screen security" feature designed to block screenshots of your Signal chats.

signal.org/blog/signal-...
By Default, Signal Doesn't Recall
Signal Desktop now includes support for a new “Screen security” setting that is designed to help prevent your own computer from capturing screenshots of your Signal chats on Windows. This setting is a...
signal.org
May 21, 2025 at 4:46 PM