@kellogh@hachyderm.io
@kellogh@hachyderm.io avatar

kellogh

@kellogh@hachyderm.io

I'm a software engineer and sometimes manager. Currently #Raleigh but also #Seattle. Building ML platform for a healthcare startup. Previously, built an IoT platform for one of "those" companies.

Open source: dura, fossil, Jump-Location, Moq.AutoMock, others

Do I have other interests? No, but I do have kids and they have interests. I think that counts for something. I can braid hair and hunt unicorns!

I put the #rust in frustrate

He/Him

#metal #science #python

This profile is from a federated server and may be incomplete. Browse more on the original instance.

jjimenezshaw, to programming
@jjimenezshaw@mapstodon.space avatar

"If the authors of computer programming books wrote arithmetic textbooks..."

A colleague used this image to describe the documentation of a library. Meaning that the documentation was the example with the rabbits, but they have to use the library as the second part of the image.

kellogh,
@kellogh@hachyderm.io avatar

@futurebird @jjimenezshaw right, its CS papers that do this

mekkaokereke, (edited ) to random
@mekkaokereke@hachyderm.io avatar

Some of my friends that are convicted felons are 100% innocent of what they were accused and convicted of. Some of my friends that are convicted felons are guilty, and absolutely did what they were accused of. I don't make fun of people for being convicted felons, and I don't ostracize them. Instead, I don't let bad people around me. My definition of bad focuses on the choices that a person made in the past, and the choices that they make now. Some of the worst people, will never be felons.

1/N

kellogh,
@kellogh@hachyderm.io avatar

@mekkaokereke yes, but when confronted with a person who has strong negative preconceived notions of felons, i will definitely make a strong association between Trump and felony, and use those biases against Trump.

OTOH, when they try to downplay felonies in order to excuse Trump, i will also definitely use that as an opportunity to educate them on the racial divide regarding felony convictions by agreeing heartily with them

lowqualityfacts, to random
@lowqualityfacts@mstdn.social avatar

I wish I could unlearn this.
https://patreon.com/lowqualityfacts

kellogh,
@kellogh@hachyderm.io avatar

@lowqualityfacts yep! right at the sternum. most heart disease is caused by toenail fungus

kellogh, to LLMs
@kellogh@hachyderm.io avatar

it’s a little disingenuous to refer to as because you can really only open source an LLM in roughly the same way you open source a microprocessor — RISCV is open source, the plans for it anyway, but it still costs millions to riff off it and make your own custom version, same with LLMs. that’s not exactly what open source was going for

kellogh,
@kellogh@hachyderm.io avatar

@wagesj45 oh wow, i get what you’re saying, but that screenshot rubs me the wrong way. nooot a good example

kellogh, to random
@kellogh@hachyderm.io avatar

it’ll be fascinating to read a hindsight analysis of the #recall debacle in 5 years. i think it represents a significant business failure, maybe a critical one. i know there’s a lot of undercurrents and dynamics at play, it’ll be a great case study down the road

kellogh,
@kellogh@hachyderm.io avatar

@gimulnautti yeah, it’s a totally different risk profile. i wrote about it here https://timkellogg.me/blog/2024/01/11/application-phishing

kellogh,
@kellogh@hachyderm.io avatar

the reason i think it’s interesting is that windows isn’t important to microsoft, and hasn’t been so for a while. and this move is in line with their general strategy — move from software to SaaS. and also more current, their deep and risky bets on AI. there’s so much that makes sense about #recall that i understand why it happened, but why aren’t they acknowledging and reacting to a very obvious mistake? disfunction? some other dynamic?

kellogh,
@kellogh@hachyderm.io avatar

and further — if microsoft culture is now one where something like recall is encouraged, that also means their security culture is one where something like recall isn’t red flagged. that’s probably the most telling aspect

kellogh,
@kellogh@hachyderm.io avatar

i love #LLMs and all, but #ML has always had a more difficult security profile and LLMs even more so. lots of subtle issues, and conflict with business opportunities

but that means that if you want to “invest heavily in #AI”, you HAVE TO also invest heavily in security. they go hand in hand.

if #recall is showing cracks in microsoft’s security culture, you can probably use that information to make predictions about their long term health

kellogh, to llm
@kellogh@hachyderm.io avatar

the energy cost of training an #LLM is about the same as the energy required to raise 2 kids. But unlike kids, you can copy #LLMs and amortize that energy cost across inferences

https://cacm.acm.org/blogcacm/the-energy-footprint-of-humans-and-large-language-models/

kellogh, to random
@kellogh@hachyderm.io avatar

i haven’t been given many “performance problems” talks in my career, but one notable exception was when i printed out business cards for myself with the title “Code Efficiency Officer”

the CEO (uh, the chief executive) couldn’t maintain a straight face when talking to me but i think he may have disliked them

kellogh,
@kellogh@hachyderm.io avatar

another time, at the same startup, my brother “gifted” me with a roll of toilet paper with $100 bills printed on them. the investors were coming over to the office for a board meeting and the CEO was busy making sure all the kegs were full, etc. so i did the obvious thing and installed said $100 toilet paper into the bathroom. i never owned up to that one (although he probably deduced it)

ErikJonker, to ai
@ErikJonker@mastodon.social avatar

It seems that Google is failing again with the first attempts of integrating AI into search or are we just seeing the terrible mistakes and is 98% of the experience great ?
In marketing/PR terms , things are not going well i think. For OpenAI it's easier they don't have a userbase with billions of users (Gmail, Search, Drive etc.).

kellogh,
@kellogh@hachyderm.io avatar

@ErikJonker yeah, the trustworthiness of LLMs isn’t really an issue. the problem is their profile of trustworthiness didn’t match Google’s existing product

when you create a new product, e.g. perplexity, it’s all new so any feature is positive. when you change an existing product, each feature is a diff from the previous state. so in this case, LLMs made google decidedly worse

if they had launched a totally new product, it probably would have had a mostly positive response

kellogh, to random
@kellogh@hachyderm.io avatar

i used to have a lot of trouble getting the kids to shut the outside doors, but then i started phrasing it as “to keep snakes out”, and i don’t have that problem anymore

kellogh, to random
@kellogh@hachyderm.io avatar

this wouldn’t ban fossil, afaict, because i’m basing fossil’s algorithm purely on what the author said, its not customized to the user at all. that seems like a good way to separate “good algorithms” from bad ones

“The New York legislation defines an “addictive feed” as one that recommends, selects or prioritizes media based on information associated with a user or their device.”

https://www.nbcnews.com/news/amp/rcna155470

kellogh, to random
@kellogh@hachyderm.io avatar

i’m teaching my 8yo about how the internet works

imo kids need to have an intuition for what computers are doing. if your video isn’t loading, why does shaking the phone not work? what does work? why?

in the past, kids learned about mechanical stuff like cars. now they need to know about computers, in order to live in an increasingly complex world

kellogh, to random
@kellogh@hachyderm.io avatar

software quality techniques outside of unit testing are fascinating

for example in machine learning, they use metrics like accuracy & F1 score as well as peer review, and then the solution is rewritten for production

the rewrite also needs to hit the same metrics as the prototype, so it ends up acting in the same capacity as the “three plates” phenomenon in unit testing https://timkellogg.me/blog/2022/04/11/three-plates

kellogh, to Metal
@kellogh@hachyderm.io avatar

my mom: “what is that racket? i can’t hear myself think!”

me: “wait, you want to hear yourself think??”

kellogh, to random
@kellogh@hachyderm.io avatar

y’all aren’t mischievous enough
https://hachyderm.io/@kellogh/112530741846356237

kellogh, to Metal
@kellogh@hachyderm.io avatar

hot take: "Violence Against Nature" is basically the version of being rick rolled https://www.youtube.com/watch?v=RHu6vJxS_6I

kellogh, to Metal
@kellogh@hachyderm.io avatar

yo new darko!! https://songwhip.com/darkous/starfire2024

kellogh, to random
@kellogh@hachyderm.io avatar

it is NOT IDEAL to have a song with an F bomb in the chorus stuck in your head, but it’s a banger so i’ll deal https://songwhip.com/falling-in-reverse/all-my-life

jimfl, to random
@jimfl@hachyderm.io avatar

Is there a medication that attenuates the metal on metal screeching of mangled machinery in my head when people refer to paddling as rowing, and/or paddles as oars?

kellogh,
@kellogh@hachyderm.io avatar

@jimfl absolutely fucking not

kellogh, to random
@kellogh@hachyderm.io avatar

this morning my dad was going off about how black fathers aren’t present in their kids’ lives. i managed to dispel this myth with data without even once declaring, “good lord, dad, where tf were YOU when i was growing up?”

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • JUstTest
  • mdbf
  • InstantRegret
  • ethstaker
  • magazineikmin
  • GTA5RPClips
  • rosin
  • modclub
  • Youngstown
  • ngwrru68w68
  • slotface
  • osvaldo12
  • kavyap
  • DreamBathrooms
  • Leos
  • thenastyranch
  • everett
  • cubers
  • cisconetworking
  • normalnudes
  • Durango
  • anitta
  • khanakhh
  • tacticalgear
  • tester
  • provamag3
  • megavids
  • lostlight
  • All magazines
    •