katee

Project Debug | General Engineer | Singapore (HYBRID) | Full-time | debug.com

Project Debug raises, sorts by sex, and releases millions of (non-biting) male mosquitoes in Singapore to reduce the local mosquito population and prevent the spread of dengue and other mosquito-borne illnesses. Our trial in Fresno, CA showed a 95% reduction in female mosquitoes during the seasonal peak.

Our engineering team is pretty small and we need a solid generalist engineer who is interested in becoming a domain expert in our factory, where we do automated rearing and sex sorting. If you run a homelab, have tinkered with Arduino, and can prompt/write some Python, we can teach you everything you need to know about our stack (Go, Python, React, Kubernetes, Tensorflow).

Reach out to katejm@debugproject.com and please mention HN in your email.

E11 bio recently released https://www.biorxiv.org/content/10.1101/2025.09.26.678648v1....

There’s an animated version at https://www.e11.bio/ and the dataset is publicly available: https://aws.amazon.com/blogs/publicsector/unlocking-the-brai...

That’s a big data application of expansion microscopy. I really enjoyed the charming smaller scale examples in the article.

Are you sure it was really a Chase representative on the phone with you? It’s sometimes possible to cause an account to be locked without being able to log in by doing too many login attempts.

[I know this is tangentially related to your point] Do you have a source on "leather cows"? My understanding is that cow leather is nearly entirely a by-product of beef and dairy cows.

Oh definitely! Doing high pressure practice in preparation for my last round of interviews definitely helped, and that round resulted in multiple FAANG offers.

VC interviews also seem very high pressure! I've met folks making funding decisions, but never been in the hot seat with them. It does seem closer to the "walk me through your portfolio/a project" interview than a whiteboard interview. Although obviously the stakes can be much higher.

I'm feel very lucky to have gotten my first "programming job" without a technical interview (it was an unpaid high-school internship that I turned into a programming job, and eventually became full-time after I was old enough to be an employee). Interviews are still tough for me, I have experienced interview anxiety extreme enough that I went home instead of continuing a day of on-sites. The interview anxiety is totally unlike anything I deal with on the job, including during other "high stress" activities like presentations, or pair coding and debugging.

Starting my career I stepped into a totally different world. I have a lot of empathy for the situations Austen that mentions here, which are mostly even more stark than my own. I also have a ton of embarrassing stories and memories. Moving between totally different socio-economic contexts is difficult, and complicated.

Tools like this are extremely necessary and important. I think that platforms (Twitter, Reddit, etc) don't have a real incentive to build out powerful tools to stop this stuff. If you aren't detecting that comments are noxious garbage, then those comments just look like engagement. It's much harder for a platform to notice/quantify the loss from folks leaving or not engaging because of the bad behavior they see, not to mention the direct harm to folks receiving the hate.

When I emailed Daniel (dang) about it ages ago it sounded like there wasn't much automation in moderation on HN. If HN doesn't have the tools, then I expect random Discord, subreddit, etc moderators are also doing their work by having humans wade through manually.

I think Tracy Chou / Block Party are more likely to succeed than the platforms are, since in this case the incentives are actually aligned.

test

Huh, I've tested on a bunch of devices/connections and haven't encountered that. Do you know what causes AMP to be that slow for you? I'll take a look at serving non-AMP pages by default. It will require tweaking how image inclusion works.

You can't make a valid recursive tree without a pre-image attack against SHA1. However `git` doesn't actually verify the SHA1s when it does most commands. If you make a recursive tree and try `git status` it will segfault because the directory walking gets stuck in infinite recursion.

This is super neat! If it weren't for the FAT limitations you could make one of those SD cards that fake their capacity actually "contain" a file as big as the misreported capacity.

I'm glad you and squeaky-clean wrote these comments. When I was experimenting in the Python REPL, I was confused by the last line here:

    >>> 100 is 100
    True
    >>> (10 ** 2) is (10 ** 2)
    True
    >>> (10 ** 3) is (10 ** 3)
    False
    >>> 1000 is 1000
    True

That's really cool! I used the web audio stuff to get real time audio data for interactive visualizations when I did a 5 minute demo of quietnet, the code is available (but might be a pain to get running): https://github.com/katee/quietnet-presentation

Running fully in the browser with visualizations might make a good teaching tool for digital signal processing (although you'd need better DSP code to demo, mine is not a shining example.)

It actually works much better when sending an receiving on the same computer. It's not as fun though.

It's definitely a toy and a bit dodgy. I wrote up some instructions over here: https://github.com/Katee/quietnet/issues/3#issuecomment-3187...

Basically I only every worked on this until it was at the point where it was working for me on the laptops I was using, it's definitely not robust.

Definitely in no way a unique idea. Also that article is awesome.