kuzushi
banner
kuzushi.bsky.social
kuzushi
@kuzushi.bsky.social
I am the co-founder of @cactuscon | ex @bishopfox ex @spiderlabs | currently pursuing a phd at UdeG in ML & OffSec. https://sensecurity.io

slayer of applications | not a super villain
Pinned
The handle "kuzushi" comes from Japanese martial arts (most frequently Judo), specifically around the idea of destabilizing something or taking it off balance. I started using it online in the early 2000s, and a friend told me it sounded like a supervillain name. Hence, the clarification.
my computer has started to act questionably. I _hate_ upgrading windows operating systems from one version to another-- it seems like it _always_ is cleaner to just do a fresh install after backing up data. Anyone ever have luck with this?
April 27, 2025 at 3:15 PM
I think my new unfollow criteria will be people who recommend books. I have to protect my wallet somehow.
April 9, 2025 at 3:27 PM
Just published a new post on setting up my local LLM environment and why. First step in the right direction :-)

www.sensecurity.io/hollme-labs/
HoLLMe Labs
TL;DR: I built a local LLM lab with Ollama + Open WebUI + ChromaDB, integrated into VS Code using Continue. Here’s how I set it up, and why it works better (and cheaper) for my heavy experimentation t...
www.sensecurity.io
April 1, 2025 at 5:06 AM
I wish I had more statistic friends. Is there a general tool where I can send a dataframe/csv tool to work with directly and start building general data explorations with that people use? right now I am doing everything by python/hand--lol.
March 27, 2025 at 8:37 PM
vibe coding is a dirty word it seems. But oddly enough, I seem to be able to make mvp projects really fast and I know how to fix them. I dont know why everyone complains so much.
March 26, 2025 at 12:43 AM
Reposted by kuzushi
Working on the talks for Microsoft BUILD and I proposed that the sessions for Microsoft Quantum be both rejected and accepted at the same time. No one gets my humor.
on a meeting trying to convince them to let us do a cool talk at Microsoft Build - "stop talking scott" was the feedback
March 21, 2025 at 7:23 PM
I am lately becoming more and more a fan of promptless ai. That is, the ai itself is hidden from the end user, but supports bridging fuzzy gaps in the systems we work with. context could support routing, just like a 10x more flexible provider model
March 12, 2025 at 7:56 PM
real talk: have you ever been convinced you were wrong by an argument on the internet? if so, how?
March 11, 2025 at 4:40 PM
I just want to ask-- do people actually talk cybersecurity and or artificial intelligence here? most of my feed feels like people talking politics and news-- which is fine. just... not exactly why I joined.
March 10, 2025 at 10:04 PM
where can I get one! haha take my money!
$14k open source humanoid robot upper torso. Writing with a pen on a notebook that you're holding is an impressively challenging task! Also comes with an open, modular, python software stack for robot control and planning.

openpyro-a1.github.io
March 10, 2025 at 9:02 PM
manus looks interesting, but basically it seems to just excel at making up for poor reasoning of the person using it. If you watch, it's approach is simply to "fill-in" the gaps of the prompt, but I am not sure that is a good thing.
March 10, 2025 at 7:07 PM
Reposted by kuzushi
March 7, 2025 at 1:07 PM
who would have thought that little touch to let you hang your backpack/laptop bag off your luggage would be so useful...
March 6, 2025 at 3:29 AM
So many great things going on in my mini-world of research. I just need to like, actually get time to do them. lol.
February 19, 2025 at 2:44 PM
American politics is like watching a train wreck of innocent people crashing into a building full of folks who are guilty. I am not sure how I am supposed to feel about anything...
February 7, 2025 at 3:05 PM
The theme I hear/see from many people is that it is harder to break into the security industry than it is to break into many/most applications.
February 6, 2025 at 6:32 PM
I think a lot about immigration lately. Watching the social medias unfold, it is a bit heart breaking. If not for birthright citizenship, i'd not even be a US citizen. No laws were broken either-- they just, changed over time. Apparently, so are we.
February 6, 2025 at 1:38 AM
I think I have a plan to automate a whole bunch of work off my plate. today is a good day.
February 4, 2025 at 4:11 PM
Trimmed the list, rebuilding again tomorrow :/
February 2, 2025 at 9:51 PM
How is excellence measured and demonstrated in your field of study?
January 18, 2025 at 6:27 PM
I wish I knew more data sciency / marketing people
January 18, 2025 at 5:42 PM
I had a teacher once say, "Nothing is more powerful than an idea whose time has come." AI's time has come, and people betting against it are going to lose.
January 18, 2025 at 5:17 PM
Frieren the anime is absolutely spectacular, and you should watch it if you get the chance. It is on Crunchyroll right now and worth every minute.
January 17, 2025 at 8:47 PM
It is interesting how Luhmann organized his Zettelkasten (slipbox). In a way, his approach is not too disimilar to vectorizing and clustering of ideas so that similiar ideas end up connected. Organizing data this way makes connections easier to work with, he just didn't have n-dimensional space.
January 12, 2025 at 8:05 PM
Perhaps it is just me, but I found myself in a pattern where deep studying of a topic (ML) actually distracted me from using it. Math, Foundations, building tests-- those are of course important. But so is like, prompting and using genai.
January 11, 2025 at 10:42 PM