Seth Kimmel
@sethkim.me
Founder @stealth data/AI infra startup
Thinks about airplanes, data, markets, math, my next meal
sethkim.me/l/
Thinks about airplanes, data, markets, math, my next meal
sethkim.me/l/
Pinned
Seth Kimmel
@sethkim.me
· Oct 28
Hi Bluesky! Quick intro post:
- Huge software/data/math nerd like many folks here
- Instrument-rated pilot
- Founder of a stealth data/AI infra startup. Will share more here soon!
- Huge software/data/math nerd like many folks here
- Instrument-rated pilot
- Founder of a stealth data/AI infra startup. Will share more here soon!
Really great collaborating with @nathanlile.bsky.social!
Reach out if you're working on synthetic data generation, offline RL, or simulating agentic behavior.
Reach out if you're working on synthetic data generation, offline RL, or simulating agentic behavior.
Earlier this year we partnered with SynthLabs (synthlabs.ai), a post-training research lab, to generate a 351 billion token synthetic dataset 10x faster and 80% cheaper.
Read more in our case study: sutro.sh/case-studies...
Read more in our case study: sutro.sh/case-studies...
SynthLabs x Sutro: Scaling and Accelerating Synthetic Data Generation for RL
By partnering with Sutro, SynthLabs generated a 351 billion-token dataset with 10x greater speed and 80% lower costs, turning complex research ideas into production-grade results without infrastructur...
sutro.sh
July 16, 2025 at 6:44 PM
Really great collaborating with @nathanlile.bsky.social!
Reach out if you're working on synthetic data generation, offline RL, or simulating agentic behavior.
Reach out if you're working on synthetic data generation, offline RL, or simulating agentic behavior.
Had a lot of fun with this one! Turns out that >1.5% of posts on Hacker News are aviation-related in recent years. www.skysight.inc/blog/hacker-...
April 3, 2025 at 11:38 PM
Had a lot of fun with this one! Turns out that >1.5% of posts on Hacker News are aviation-related in recent years. www.skysight.inc/blog/hacker-...
Gave a fun talk at the @modal-labs.bsky.social and Mistral AI
demo night last week in SF! We discussed open-source model security, applying some of the large-scale inference techniques we've been working on recently.
www.skysight.inc/blog/model-s...
demo night last week in SF! We discussed open-source model security, applying some of the large-scale inference techniques we've been working on recently.
www.skysight.inc/blog/model-s...
Model Security with Large-Scale Inference
_(Adapted from our talk at the Modal x Mistral Demo Night in San Francisco on March 6th, 2025)_
www.skysight.inc
March 13, 2025 at 7:51 PM
Gave a fun talk at the @modal-labs.bsky.social and Mistral AI
demo night last week in SF! We discussed open-source model security, applying some of the large-scale inference techniques we've been working on recently.
www.skysight.inc/blog/model-s...
demo night last week in SF! We discussed open-source model security, applying some of the large-scale inference techniques we've been working on recently.
www.skysight.inc/blog/model-s...
Do they have to nerd snipe us this bad?
17. Our advisor Martin Kleppman wrote the book “Designing Data-Intensive Systems.” Good thing too, because Bluesky is now a data-intensive system!
November 19, 2024 at 11:46 PM
Do they have to nerd snipe us this bad?
I've always wanted to like python notebooks
November 19, 2024 at 11:39 PM
I've always wanted to like python notebooks
Anyone want to make #graphBS a thing?
New post up! ✨
Exploring AT Protocol with Python to visualize the #databs social graph!
davidgasquez.com/exploring-at...
Took less than 1 hour to get the data and plot it. Amazing what you can do with open APIs and great SDKs!
Exploring AT Protocol with Python to visualize the #databs social graph!
davidgasquez.com/exploring-at...
Took less than 1 hour to get the data and plot it. Amazing what you can do with open APIs and great SDKs!
November 14, 2024 at 8:11 PM
Anyone want to make #graphBS a thing?
Depressing will be the day a developer asks me:
"What is StackOverflow?"
"What is StackOverflow?"
November 14, 2024 at 7:28 PM
Depressing will be the day a developer asks me:
"What is StackOverflow?"
"What is StackOverflow?"
It seems pretty clear to me that this thing will be as big/bigger than X in terms of sheer number of users
November 14, 2024 at 6:19 AM
It seems pretty clear to me that this thing will be as big/bigger than X in terms of sheer number of users
That being said, my guess is progress will resume when we start to generate high-quality, focused synthetic data. Sort of like forcing a human to go to the library and acquire actual knowledge instead of scrolling on social media all day
How can we be surprised that LLM scaling laws don't hold when the training data is literally just crap people write on the internet?
November 13, 2024 at 3:48 AM
That being said, my guess is progress will resume when we start to generate high-quality, focused synthetic data. Sort of like forcing a human to go to the library and acquire actual knowledge instead of scrolling on social media all day
How can we be surprised that LLM scaling laws don't hold when the training data is literally just crap people write on the internet?
November 13, 2024 at 3:46 AM
How can we be surprised that LLM scaling laws don't hold when the training data is literally just crap people write on the internet?
Btw the bait is me just saying there is no debate
Okay now that everyone is here who wants to get baited into an R vs. Python debate?
November 13, 2024 at 1:01 AM
Btw the bait is me just saying there is no debate
Okay now that everyone is here who wants to get baited into an R vs. Python debate?
November 13, 2024 at 1:01 AM
Okay now that everyone is here who wants to get baited into an R vs. Python debate?
Seems like the holiday is bringing huge amounts of new users over here. @jaz.bsky.social is your data supporting this?
November 11, 2024 at 11:05 PM
Seems like the holiday is bringing huge amounts of new users over here. @jaz.bsky.social is your data supporting this?
Who is going to build the billion dollar slop/non-slop classifier?
hey bsky, feature request: a moderation option that blocks AI-generated images
threads, IG, pinterest, etc. are swimming in them, and giving users the option to exclude them could really improve and differentiate the experience here
cc @pfrazee.com
threads, IG, pinterest, etc. are swimming in them, and giving users the option to exclude them could really improve and differentiate the experience here
cc @pfrazee.com
November 11, 2024 at 5:15 AM
Who is going to build the billion dollar slop/non-slop classifier?
While a lot of the content on X is clearly written by humans, it's sort of degraded into subhuman patterns of engagement. Lots of cryptic speak, baiting, trolling, inflaming, etc. Glad this place has actual humans posting actual human thoughts.
November 11, 2024 at 5:05 AM
While a lot of the content on X is clearly written by humans, it's sort of degraded into subhuman patterns of engagement. Lots of cryptic speak, baiting, trolling, inflaming, etc. Glad this place has actual humans posting actual human thoughts.
This is super cool! One has to assume that the most open, programmable, and hackable content platforms win in the long run
November 9, 2024 at 8:44 PM
This is super cool! One has to assume that the most open, programmable, and hackable content platforms win in the long run
Becoming a more confident engineer isn't about writing less dumb code; it's about accepting the fact that everyone else's code is just as dumb as yours
November 8, 2024 at 9:54 PM
Becoming a more confident engineer isn't about writing less dumb code; it's about accepting the fact that everyone else's code is just as dumb as yours
GenAI - pay more for inefficient ML models
Crypto - pay more for others to verify your own transactions
SaaS - pay more for something a spreadsheet can do
Cloud - pay more for someone else’s computer
Mobile - pay more for apps that can run in a browser
Crypto - pay more for others to verify your own transactions
SaaS - pay more for something a spreadsheet can do
Cloud - pay more for someone else’s computer
Mobile - pay more for apps that can run in a browser
November 8, 2024 at 6:21 PM
GenAI - pay more for inefficient ML models
Crypto - pay more for others to verify your own transactions
SaaS - pay more for something a spreadsheet can do
Cloud - pay more for someone else’s computer
Mobile - pay more for apps that can run in a browser
Crypto - pay more for others to verify your own transactions
SaaS - pay more for something a spreadsheet can do
Cloud - pay more for someone else’s computer
Mobile - pay more for apps that can run in a browser
People forget that it's not unusual for Apple to release products that initially suck and are iteratively refined. They take big bets.
The original iPhone, Apple Maps, etc.
My guess is Apple Intelligence will have a dominant, frontier consumer AI product within ~5 years.
The original iPhone, Apple Maps, etc.
My guess is Apple Intelligence will have a dominant, frontier consumer AI product within ~5 years.
Apple was late to the AI game, but the release of Apple Intelligence is proving that it can destroy its reputation and become every bit the laughing stock Google and Meta became.
November 3, 2024 at 11:00 PM
People forget that it's not unusual for Apple to release products that initially suck and are iteratively refined. They take big bets.
The original iPhone, Apple Maps, etc.
My guess is Apple Intelligence will have a dominant, frontier consumer AI product within ~5 years.
The original iPhone, Apple Maps, etc.
My guess is Apple Intelligence will have a dominant, frontier consumer AI product within ~5 years.
Startup idea: Cursor, but it just shits on your patterns and bullies you into refactoring your entire codebase every time you ask it a question.
I used to work with someone who believed design patterns were some holy grail of engineering achievement and tried to fit every problem into one, often before shipping a single line of code. He was always frustrated with the lack of perfection in what others would build, and so would never ship.
November 3, 2024 at 10:53 PM
Startup idea: Cursor, but it just shits on your patterns and bullies you into refactoring your entire codebase every time you ask it a question.
What would an internet for LLM's look like? (not for human consumption) Some thoughts below:
October 31, 2024 at 7:57 PM
What would an internet for LLM's look like? (not for human consumption) Some thoughts below:
This trillion parameter LLM inference could have been a y = mx + b
if I could only give one #dataBS advice it would be to start with a simple model and use it as a benchmark that you check more complex models against
October 30, 2024 at 8:56 PM
This trillion parameter LLM inference could have been a y = mx + b
This is some serious #databs
Checking in a little bit later now and we're still on a roll, up to just about 13.5M users from ~6.2M users 2 months ago.
Good thing there's nothing exciting happening next week and things will be nice and normal and stable for a while...
Good thing there's nothing exciting happening next week and things will be nice and normal and stable for a while...
October 30, 2024 at 3:53 AM
This is some serious #databs
