
I work on LLMs, security, and distributed systems. CTO at @babylonlabs_io.
Joined August 2013
- Tweets238
- Following492
- Followers504
- Likes2.1K
AI is eliminating the labor-cost barrier that previously provided an accidental layer of security to most of the internet
1
67
Great way to deslopify your feed
Weekend hack: HackerTube. I was annoyed with YouTube's recommendations so I had Claude filter down 43M comments over 20 years of HN to 6,454 videos that keep being recommended. *Significantly* better viewing experience, ~2 hours with Opus 5.5:
hackertube.org/
- Excellent recs without any logins (state is kept in local browser session)
- Thumbnail is a representative frame from middle of the video instead of the super duper A/B test hyper-optimized user uploads
- Underlying data from the comment analysis as one zip file, point your Claude at it and ask for your own version
- HackerTube Longs ✊
A bit more on the process:
I use Claude Code (desktop app for Mac), Opus 5.5 on Medium effort as my default. I only talk to Claude and have it do everything, the only "manual work" I helped with in tandem was in the browser (reminiscent of MenuGen): 1) I had to create some kind of API key from Google Cloud UI to query YouTube for video metadata, 2) I booted up a VPS box for serving and 3) I bought the domain name. Once the box was up I gave Claude the IP and told it update, secure, harden, deploy and optimize the code, and it did the whole thing.
Many awesome projects that would have normally taken weeks/months even can now be few-shotted for breakfast. What else are we not thinking of.
49
Vitalis Salis retweeted
Surprise 1: Anthropic acquired a one-person biotech
.
.
.
Surprise 2: that person is me
Surprise 3: I’m 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
…and yes, it has gotten so boring around here that I’m 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, ‘elsewhere’ 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, I’m grateful for it, & I'd partner with Atlas again in a heartbeat.
In typical AV fashion, Mithril’s 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 “Pipeline-in-a-Person.” The idea immediately resonated when we shared it here back in April; and it was central to the shared vision that led to Anthropic’s 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, imo… 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 I’m willing to bet they’ll 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 anyone’s earned a proper introduction, you all have.
It’s 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 I’ve been able to pay it forward, in my own small way.
But, to be clear, Prof Oak isn’t 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
Vitalis Salis retweeted
We are out of stealth! As AI writes more code, we're building tools to prove programs meet precise requirements, with Lean proofs anyone can independently check. Let's make verified software the norm in every language and application.
Come take a look: andtruth.com
Vitalis Salis retweeted
Two weeks ago a life changing amount of money was wired into my bank account from the sale of my business.
Since I'm no longer a competitor in this particular arena, I'm gonna drop some sauce on how I completely mogged an entire industry on meta ads.
I'm going to be vague so I don't dox myself, but here it goes:
it's great to see the security community stepping up with more defensive tools
today we're releasing apex-flash-1, our first open-weights model for security research post-trained on real vulnerabilities we found and got paid for.
we are also releasing the abliterated variant for researchers who want fewer refusals in their own authorized workflows.
how we built it: cantina.security/apex-flash
5
118
just one of the implications of everyone having a smart personal assistant in their pocket
Fascinating.
Chief Economist at Apollo: agents could cause a bank run by sweeping household cash into accounts paying 3-5% instead of the 0.1% national average, causing banks to lose a large share of their cheap deposits.
1
118
There’s a major trap in being over-optimistic about timelines: yes, the tough target makes you move faster, but you also make risky commitments that will probably not happen. Quality might also suffer if there’s not the discipline or necessary systems in place
1
1
74
Great to see a competition format that fosters collaboration. Having access to the latest improvement as soon as they get made allows for optimisation to happen at a significantly faster pace
Introducing @YukonResearch
A platform for open frontier research.
Over the past 2 months, an open network of humans + AI has already:
- Beat Google’s frontier quantum circuit result over 50%
- Made Poolside’s open-weight model run 2.6x faster
- Increased post-quantum Ethereum scaling by 3.5x
- Boosted Lighter’s prover 9.5x faster
Put a hard frontier science problem in front of many independent solvers, each bringing different models, harnesses, prompts and tacit knowledge. And let the best results compound.
Accelerating the future of scientific progress together with AI.
Bring us your hardest problem: yukon.org
Access to unique data in your field means you can post-train cheaper models optimised for that field. It’ll be interesting to see which incumbents use this advantage to stay on top in the ai era, and which get toppled.
57
Great points here
1) The right time to be building a harness was 12 months ago. If you're building one now, you're too late.
2) The right thing to do today is to own the entire loop. From the model to the harness to the data to the right context. Every iteration needs to make the loop better, faster, and cheaper.
3) The thing to optimize for is intelligence per dollar. You can be SOTA in a domain, both on capability and cost.
4) This is a bit of a secret. Security saturated in January 2026. Anything you can do today, you can do with frontier models from January 2026. Everything else can be replicated with test-time compute! The challenge today is how to achieve a certain outcome for the lowest cost -- a Pareto-optimal product.
5) A simple coding harness and a goal will work, but it is not Pareto-optimal. It's far from it. If you know what you're doing, you can get 10-100x efficiency per dollar for the same outcome.
6) Cyber is increasingly getting restricted, and it'll get even more restricted. We see this in our benchmarks. Even with trusted access, there are several places where cyber is still lobotomized.
7) Don't believe in the story the big labs are telling you. The recipe is all the same. To make Claude better in a domain, Anthropic needs domain-specific knowledge. You either give it to them for free on the internet, or get paid, or keep it to yourself for your post-training. It's rumored Anthropic's margin on the API is 90%+. This is why post-training OSS with your own domain specific data and owning the outcome is what application companies are chasing right now.
8) Scale is underrated. Tokens are this infinite lever if you can scale it. We did around 5 trillion tokens in the first six months of the year. Now we're building a trillion token a week pipeline. That much on Claude would be hundreds of millions a year. It's possible to achieve that for 1/100th of the cost. A human equivalent of that labor would be tens of billions of dollars!
7) Lastly, there are no hard rules here. This is the most dynamic time in tech, where you need to constantly check your assumptions and be willing to adapt almost on a daily basis. The key thing to think about is that the intelligence per dollar for nearly all tasks is falling every day. What can be built today that wasn't possible before (where is it on the cost curve), what unique data do you have that lets you be Pareto-optimal on a domain? Teams used to build products with one core, often technical idea, and build month- or year-long roadmaps around it. That is straight up not going to work. The only moats you have are building an adaptable team, owning unique data, and unique domain-specific flywheels. Some of the players are going all the way through, including finding cheap energy and custom chips to run inference. It's an intense period in time, and the biggest risk you could be taking here is sitting out of the game.
2
93
Vitalis Salis retweeted
The Vault is back with the new host @jk_remote sits down with Babylon's CTO @vitsalis.
We are unpacking:
- why using native Bitcoin as collateral has been such a difficult technical problem
- what trade-offs wrapped and bridged Bitcoin have
- how TBV keeps native Bitcoin while making it usable as collateral on Ethereum.
Watch the full episode👇
Situational Awareness, the $20bn hedge fund founded by former OpenAI employee Leopold Aschenbrenner, has sought to raise fresh capital from investors after suffering heavy losses during the recent rout in AI stocks ft.trib.al/SwCjjE6
Lots of talk about web3 security in the age of AI at WebX. We need to be talking more about setting up some common standards and evals. The AI auditing space still feels like the wild west with everyone building their own thing.
1
120





















