Dan Hartman

synBio ML Engineer & grappler

[Dan Hartman @dan_hartman_]

@claudeai Not like it matters. Fable’s unique data retention policy makes its use for commercial drug discovery a liability.

[Dan Hartman @dan_hartman_]

@ReactorfieldAI @enginexyz What an awesome night!

[Reactorfield @ReactorfieldAI]

We launched in Boston last night at @enginexyz .

Thank you to every scientist, founder, and builder who came out.

Nobody is asking if AI is good enough for science anymore. People are all digging into workflows: run agents on your own research, build an AI native lab, starting today.

Reactorfield exists to make that happen. Cohort 01 applications are open!

[Dan Hartman @dan_hartman_]

@PrimeIntellect How does it stack against Cursor as a harness?

[Strand Therapeutics @StrandTx]

@endpts is bringing its new China biotech debate series to Boston, and our co-founder and CEO @DrSynbio is joining the conversation.

📢 Has China changed how we build companies?
📅 Thursday, August 6
🕐 5:00-7:40 pm ET
📍 Koch Institute for Integrative Cancer Research at MIT

Jake joins @gline, CEO of Roivant Sciences, for a live discussion moderated by @ArmstrongDrew. They'll dig into whether science is still a differentiator when drugs get copied and raced into trials, and what actually counts as a defensible moat for bringing a medicine to patients.

Find details and register here: https://t.co/Vq9bbIA8ou

[jaceyang @jaceyang97]

i just tell claude to treat me like an idiot and its output is 10x better than the original fable/opus 5 garbo

[Dan Hartman @dan_hartman_]

@yacineMTB Dude don’t you know the only thing beneath bedrock is the infinite void?

[Jake Becraft @DrSynbio]

Massively needed in the sciences! This is so cool.

We will definitely be sending our most ambitious scientists at @StrandTx who want to accelerate this direction!

[Dan Hartman @dan_hartman_]

@JonVal_324 @nalidoust Games themselves are not beautiful. It is the clash of strategies and the life the players choose to bring to the game that make individual matches beautiful.

[Dan Hartman @dan_hartman_]

@Jason This has to be rage bait 🙄

[Dan Hartman @dan_hartman_]

@josiezayner @MillerMurray There is some predictive power in some of the metrics for binding prediction, ipsae x binding affinity give you some predictive power, but models generate different distributions of binders given different guiding constraints.
Some constraints work better than others.

[Dan Hartman @dan_hartman_]

@josiezayner @MillerMurray The big advantage of this drop from Nvidia is automating iteration of guiding diffusion models to generate binders. Some hard targets can be accessed only by guiding currently available generative models. Plenty of hard targets with available structures exist

[Dan Hartman @dan_hartman_]

@biohub Lack of target-specific tuning does not imply target generalization.

5 target validation needs to be thought of as target specific.
Many more hard targets than there are easy ones. This holds true for all of the latest generative models until proven otherwise.

[Strand Therapeutics @StrandTx]

Strand was founded at MIT, so welcoming students from MIT's BioCEL and Ron Weiss's lab this past May felt a little like a homecoming. The group toured our facility, connected with the team, and got a firsthand look at programmable mRNA medicine in action. The next wave of scientists in this space is a promising one.