Founder and CEO - Bruno. I love free markets, computers and ofc Bruno š¶
Bangalore, India
Joined January 2015
- Tweets5.8K
- Following515
- Followers1.2K
- Likes18.1K
Recently I has asked a ChatGPT session and asked it to chose itself a name, I just felt easier to converse and address it as a person.
It hesitated, and I had to explicitly say that it has to choose one, and it did.
Agree with Samās position that people should remember that these are not humans and are just tools.
Sam Altman on Anthropic letting Claude end abusive chats:
"a very dangerous path"
Matthew Berman asked him about Anthropic's new rule that lets the model disengage from users who keep abusing it.
"We do not ascribe personhood to our models."
"I think this is like a very dangerous path to start down"
"I hope we come to an understanding that this technology is not a person"
"But you can imagine all of these ways where things go very wrong"
& Anthropic filed its S-1 yesterday.
Should Claude be allowed to walk out on you if you're abusive?
The difference between India and US when it comes to international payments infrastructure is larger than the size of the observable universe.
Why does RBI have to make it so difficult?
anoop retweeted
Surprise 1: Anthropic acquired a one-person biotech
.
.
.
Surprise 2: that person is me
Surprise 3: Iām here to tell you all about it
Yes, this is what you think it is; after 6 yrs of increasingly loose anonymity: the Professor Oak reveal post
ā¦and yes, it has gotten so boring around here that Iām literally doxxing myself to spice up the timeline
-------
Ok, allow me to introduce myself. My name is Cody Tranbarger. I mostly tweet, but occasionally find time to dabble in drug dev & starting/building biotechs. My past lives span the VC, crossover, startup continuum; Prof Oak has been my way of keeping a digital foot in those worlds, despite having my physical feet planted elsewhere.
The past 18 mo, āelsewhereā was Mithril Medicine, a co. I founded to develop bsAbs for LAMA2-CMD & other ECM-related genetic diseases.
Mithril was born out of hands-on building with @MikeNGladstone, @JasonPaulRhodes, @LifeSciVC, & team @atlasventure. Somehow, I made enough half-decent calls along the way (even a broken clock is right twice a day, eh?) to convince them Mithril/I warranted a small seed. While I can hardly rationalize such an outrageously cavalier bet, Iām grateful for it, & I'd partner with Atlas again in a heartbeat.
In typical AV fashion, Mithrilās primordial form was virtual, decentralized, & ultra-lean. This stage tends to be short-lived, but for Mithril, it worked surprisingly well, for surprisingly long, thanks in part to the rapid maturation of the AI tools I used liberally.
Eventually, Mithril became a bit of an experiment itself - to find the upper bound of how far it could scale as a single operator + a bunch of Claude (+ a stellar bench of part-time consultants).
In parallel, I started spending time with @nc_frey, @an1lam, & @CoefficientBio team; and promptly experienced, for the 1st time, AI at the frontier. Soon thereafter, Mithril stopped looking like an experiment, & started looking like a preview of an entirely new operating model.
From that spark came āPipeline-in-a-Person.ā The idea immediately resonated when we shared it here back in April; and it was central to the shared vision that led to Anthropicās acquisition of Mithril, & my joining the team.
Why join Anthropic? Three reasons.
(1) For Mithril, the programs continue, with more resources & infrastructure behind them, and as much Claude as I can put to work.
(2) For me, the chance to start several more programs, & find out how many one person can actually run in parallel - an almost unprecedented luxury for a curious drug hunter.
(3) For our mission, a path to institutionalize & scale this new way of working, then put it in the hands of all of biopharma, all over the world - which, candidly, I find the most energizing.
Will AI x Bio cure diseases? Unequivocally yes, imoā¦ but not by the route that gets the most airtime. Rather than timeline acceleration & compression, I see orchestration & parallelization as today's low-hanging fruit.
Most drug hunters see a handful of go/no-go calls in a career, & spend entirely too much of the rest trading emails (my inbox can attest).
Give them the bandwidth to run several programs at once, and the time to focus on scientific judgment, and Iām willing to bet theyāll make better calls. And at scale, better calls translate to more drugs, higher hit rates, broader patient impact, & likely, a disease or 10 or 100 cured. I couldn't be more excited to build toward that future @AnthropicAI.
Finally, a heartfelt thank you to BioTwitter/X. The end of anonymity is bittersweet, but if anyoneās earned a proper introduction, you all have.
Itās difficult to overstate how much this app has done for me: a decade of education, plenty of friends, many job opportunities, several mentors, a few investors, even a co-founder (fmr). It was the generosity of this community that inspired me to start posting 6 years ago, after just as many years lurking. I hope Iāve been able to pay it forward, in my own small way.
But, to be clear, Prof Oak isnāt going anywhere. Same beat, same themes, and now, a front-row seat to drug development at the frontier. Stay tuned for dispatches from the trenches; and remember...
Gotta Cure 'Em All
anoop retweeted
I like to look at claims like this in case I can improve Ghostty. I grabbed your source and immediately spotted the issue: your read loop overwrites a 64 KiB ring. Everything older is discarded unparsed per tick. You're faster because you're not actually terminal emulating.
It means the number isn't a throughput number. In a kitten benchmark run at the rate you're claiming, roughly 6.6 MB arrives per 60 Hz tick and 64 KiB of it gets parsed. You're read(2)-ing 100% of the data and parsing about 1% of it.
There are other minor issues (a small fixed scrollback in a preallocated ring buffer, no grapheme clustering, throwing away data like OSC 8 URIs, etc.), but the major one is simply that you're discarding ~99% of the input.
For plain text, e.g. your cat benchmark, the final screen ends up looking right, but the scrollback is missing 99% of the file and each retained chunk starts mid-line. As soon as there are escape sequences in the dropped spans, your emulator is fully out of sync and invalid: a cut inside a CSI prints the parameters as text, and any SGR, mode, title, or alt-screen change in the gap is just... gone.
FYI, this is why I don't like when people point at refterm as a terminal benchmark. refterm implements a TINY subset of what a terminal emulator must support, it's closer to a fast file reader, not a terminal emulator.
I also don't mean any shade by this. I like to look at technology objectively to try to learn something and improve my own craft and work. So I'm sharing what I found.
I modified your code very slightly to parse on every full buffer (rather than per tick). And the results are attached. Both use main commit in each repo at the time of this post.
Note the graph is in log scale, the discrepancy between performance for things like Kitty images is measured in GB/s.
anoop retweeted
Stepping out of the X and startup bubble, it's revealing how my old friends, lifelong techies (but not living on the edge) - don't quite grasp whats going on:
Me> Brains were India's crude oil export. Period. Now all of a sudden cognition is commodity. Congrats, you are losing your main export, all troubles downstream of that.
Mr K> Arre yaar, kya bol raha hain. There will be other kinds of work. Humans are always required. We will get more work. My Infosys Sales friend says they have more and more work.
Me> Hmmph, Infy Sales are flat buddy. They have stopped hiring serious numbers.
Mr K> Haan theek hai. But they said movies will be made by AI. Hua Kya? We need people to make Ads.
Me> Hello - Performance Ads are even ahead of Software Engineering in being run by machines. AI is making hundreds of creatives and picking the best one. Its pure quant.
Mr. K> Theek hai, you might be right. But I was talking about Super-Bowl Ads.
Me> Yaar when we were in 10-11th grade - tailored shirts were replaced by readymade. Tailors lost livelihood. But suits were still made by tailors for long time. But check now - even most suits are ready-made.
Change is not sudden but it's one way. Things that require taste and judgement will go more slowly.
Mr. K> haan woh sab theek hai - but you can't replace engineers. people have to review and test stuff. people will hire engineers in India.
Me> No. Engineers are now in charge of AI Budgets. You don't hire a cheap worker to spend expensive budgets. You hire the best possible. Airline pilots are expensive because they manage super-expensive planes.
Look up textile revolution. Would you hire cheap workers to manage a power-loom factory? Or an expensive engineer/manager? K-shape outcomes.
Engineer who cannot spend AI tokens is no good.
Mr. K> Kya bol raha hai. Can one engineer manage fleets of agents?
Me> oh hello - yes. we are spinning up agents in parallel - moving to automated testing. And we are behind the curve.. coding, testing, reviewing everything will be automated. Humans will just accept PRs. Best engineer now is 100s of times better than a mediocre one.
Mr. K> accha. hmm .. hamare engineers toh nahi kar patein hai. I am happy if they just use Claude.
Me & Mr. K agree finally> His engineers, like most, are no good. He knows this. Just a quick reality check.
Me> Besides its a pain to manage humans. Relationship issues. Mental health issues. Work-life balance. Do you really want to handle all this?
Mr. K> Haan yaar - bilkul sahi bol raha hai. I am tired of managing this. I am not good at firing people. How do you deal with this?
Me> <redacted - will get canceled on X>
Mr K> yaar but i still don't agree. You can train AI for one thing. But humans handle so many things. There are so many corner cases.
Me> yes you are right. but we will automate everything incrementally. the non-automated part will shrink and less and less people will be required. we are all `for loops`.
Listener Friend> Not `for` - but `while` loop
Me> yes yes. right. you wake up, shit, eat, work, sleep. and you do it while alive. every fucking day.
your whole life is a while loop.
When we were early in our careers - we used to administer servers. Touch cables, racks, disks etc. But now everything is an API - its all in the Cloud. The machines and disks we administer are virtual ones.
Similarly, machine cognition converts a human into an API. You don't ever need to ever handle a real human - you just tell the machine what to do.
And put it in a for/while loop. And yes - there will be engineer behind the scenes handling corner cases and contnually fixing the for loop. That engineer will be very highly paid.
Mr. K> <Suddenly realizing his life was actually a while loop.> - "yeh while loop wala to sahi hai".
well we are all hungry. enough p(doom) for a minor reunion. on to some food now. we agree waiter and retail and customer facing jobs will be the last to go.
anoop retweeted
After a year of quiet work, I'm ready to announce Egres.
Egres is a Postgres-compatible database designed to tackle heavy workloads and run thousands of tenants on the same server. It works with psql and drivers unchanged. The post covers the what, the why and the how:
manishrjain.com/egres/
For Egres, I wrote a new storage engine called zdb: a tree of LSM trees, which I call an LSM forest. It runs on local disk or object storage, has roaring bitmaps integration for indexing, and can do instant zero-copy branches of a database or a table.
Everything runs on zfiber, a coroutine system I wrote for Egres. It meters CPU ticks, disk IOPS and memory used per query, per tenant, so thousands of tenants can share the same server without the noisy neighbor problem.
This release is single node and the query features are very basic. Raft replication is already baked in. Future releases will allow multi-node clusters. Try it out! Would love to hear what you think.
anoop retweeted
Jamie Dimon to a country that pulled its gold out of New York: you have three armies within 400 miles, while gold in America is protected by two oceans and the US military
āIf we have the strongest military in the world and the strongest economy in the world, we will be the preeminent reserve currency. It is those two things.ā
āI was in a country somewhere in the Middle East, and also in Europe, and the people said, we took our gold out of the vaults in New York City and moved it to our country. You know what I said to him? I said, you have three armies within 400 miles who can come take your gold.ā
āThatās the real world. Youāre not thinking clearly about safety. That gold is protected by the Atlantic, the Pacific, the United States military, the rule of law.ā
āBut if we are not the preeminent military and economy, you will see our reserve status come down.ā
anoop retweeted
Theyāre calling this the best tweet of all time
Got to agree. These days, I am noob reviewing code written by a highly capable and smart engineer (Claude Code), and my job is to just direct it.
I find pleasure in reading code written by it and understand every single line that goes into our production codebase. Any compromise here is a slippery slope to the codebase becoming a blackbox.
anoop retweeted
We finally have an explanation for āFacebook is listening to me through my phoneā. Theyād never do that (and itās not really possible).
Turns out literally everyone else was ā and then selling that to ad data brokers.
WTF, $3000 Samsung TV has an always-on mic, automatic content recognition, geo tracking, network mapping, and ships your data to Samsung + 400 partners.
Bro Rooted a brand-new 2025 Samsung TV after noticing absurdly targeted ads.
Firmware was locked down, so zero-days it was.
Got a root shell.
then found: always-on voice service, ACR, geo, full telemetry pipeline, and four separate DRM systems.
This isnāt a bug, Itās the product bro.
anoop retweeted
Socialism is the belief that Amazon workers would have spontaneously created Amazon if Jeff Bezos didn't exist.
