Sabitlenmiş Tweet
Introducing Biomemex
An AI wet-lab copilot
Automatically track your experiments + catch mistakes
No more ‘did i aspirate that well’ or ‘why is my cell line contaminated’
Built under 24h with @ahadj0, Will, and Crystal
English
Daniel George
497 posts
Daniel George
@degtrdg
i like to build and write. prev co-founder @speck_ai (yc w24), eng @sphinx_bio
San Francisco, CA Katılım Ağustos 2018
197 Takip Edilen387 Takipçiler
Just realized it’s already been a month since I joined @thinkymachines. Time flies fast here and I’ve been enjoying life, happier and more motivated than ever to work. Come join us!
English
gork set this up no mistakes
x.com/nikitabier/sta…
Nikita Bier@nikitabier
There’s a better way: The optimal tax structure for a small business is an S-Corp. S-Corps can deduct payroll via the QBI deduction. And if you’re in NY or CA, you’re able to deduct state tax payments (and bypass SALT deduction limits) via a program called PTET. So yes, you can pay yourself via payroll and have it be tax efficient.
English
SF is the only place where you can up with an 8sleep and AirPods Maxs before you have a room bigger than a closet
English
@raveeshbhalla @Claude i wish there was a way to sync gdocs to obsidian or publish obsidian to gdoc in a way that i can have gdoc comments from others to my md files
English
current random side project: collabing with @claude on a spec for how to do comments/suggested edits in markdown files using files itself, so that coding agents can comment and do the work too
English
@otta156 we get it otta lamps are doing well. dinner’s on you then?
English
my friend misheard "effective altruists" as "cautious optimists" and i don't have the heart to correct him
English
@myhandle @maebichka i still want something with a moral dilemma not just a pure “are you agentic”. maybe a “would you switch out my order with someone else’s”.
English
I was on a dinner date, and i ordered the poke bowl but when it arrived they hadn't really cut the salmon right, the cubes were all connected. waitress was busy so i walked into the kitchen, asked chef for a knife, redid it, and my date called me BRAVE🤬
x.com/bashu_thanks/s…
bashu, going to vibecamp, thanks@bashu_thanks
what was the last thing that made you really indignant
English
@myhandle @maebichka ive used 'would you steal the scarf' in more conversations than i'd like to admit because of you. need to find a new thought experiment.
English
@maebichka do you object to "brave" because walking into the kitchen wasn't scary at all for you, and the fact that the guy thought it would be means he's a low agency normie who would never steal a scarf for you and that gave you the ick?
English
@tombielecki he doesn't have an anime pfp but ill tell him about this so that he can up his price
English
my friend with 'engineer physiognomy' just got paid 3k to sit in an office during a vc visit. he's getting flown across the country, five star hotel, free airpods max, and company card. why am i employed.
English
@ryanmart3n @karpathy big part of what im doing is having claude update my agent's scaffolding based on failures in my evals. would love to discuss!
English
Evals are the verifier for the agent building process.
Once you have an eval, you can autonomously hill climb it with Meta-Harness (agents building agents).
@karpathy’s autoresearch (agent building models) is another piece.
To complete the full autonomous loop of agent development we need Meta-Measure / autobenchmark (agents building evaluations).
We have started automating the task discrimination (quality control) part of the benchmark development process (in our efforts for Terminal-Bench 3.0), but the real unlock will be cracking task generation.
DM me if you are interested in autobenchmarking and let’s jam on some ideas!
Yoonho Lee@yoonholeee
How can we autonomously improve LLM harnesses on problems humans are actively working on? Doing so requires solving a hard, long-horizon credit-assignment problem over all prior code, traces, and scores. Announcing Meta-Harness: a method for optimizing harnesses end-to-end
English
@alexeyguzey i think hat guys should be able to challenge other hat guys for the chance to do a swap. i want to make america emo again once in a while.
English
i need more hat guys in my life @alexeyguzey
(full head of hair hat wearers only who do it for the love of the game)
English
the average friend group of 6 produces ~18 failed talking stages a year. i did the calculations and if every friend group ran a referral system we'd solve the fertility crisis, save social security, and collapse the therapy industry within a generation.
English
@kanishkg2010 when am i paying you 3k to come hang out in the next couple of months
English
