David
@davedangond.bsky.social
Software Engineer
B.S. Bioengineering @ucberkeley
Author of sprintome.substack.com
Opinions are my own
B.S. Bioengineering @ucberkeley
Author of sprintome.substack.com
Opinions are my own
My latest write up explores the evolution of AI in scientific discovery from the 60s to today.
With breakdowns of (IMO) some interesting key players in the field (@futurehousesf.bsky.social, @lilasciences.bsky.social, Biomni)
~20 min read
Lmk what you think🫳
sprintome.substack.com/p/the-evolut...
With breakdowns of (IMO) some interesting key players in the field (@futurehousesf.bsky.social, @lilasciences.bsky.social, Biomni)
~20 min read
Lmk what you think🫳
sprintome.substack.com/p/the-evolut...
The Evolution of AI in Scientific Discovery: A New Era in the Life Sciences
AI is becoming a powerful partner in scientific research, especially in bio and medicine.
sprintome.substack.com
August 1, 2025 at 10:54 AM
My latest write up explores the evolution of AI in scientific discovery from the 60s to today.
With breakdowns of (IMO) some interesting key players in the field (@futurehousesf.bsky.social, @lilasciences.bsky.social, Biomni)
~20 min read
Lmk what you think🫳
sprintome.substack.com/p/the-evolut...
With breakdowns of (IMO) some interesting key players in the field (@futurehousesf.bsky.social, @lilasciences.bsky.social, Biomni)
~20 min read
Lmk what you think🫳
sprintome.substack.com/p/the-evolut...
Reposted by David
Increasing model size from 339M to 46B parameters enables models to grasp core 'protein grammar,' leading to higher-quality sequences that pass more rigorous filters.
Size matters! More parameters = better foundational understanding in protein design.
Size matters! More parameters = better foundational understanding in protein design.
New Post: “The Protein Language Model: Bigger = Better?” — a quick dive into ProGen 3, scaling laws & alignment magic.
Massive kudos to the crew at ProfluentBio for turning AI horsepower into real proteins. 🧬🤖
open.substack.com/pub/sprintom...
Massive kudos to the crew at ProfluentBio for turning AI horsepower into real proteins. 🧬🤖
open.substack.com/pub/sprintom...
The Protein Language Model: Bigger = Better?
An Intro to Protein Language Models (PLMs) 🧬 + Scaling Laws through the lens of ProGen3: architecture, protein generation/viability, alignment. (In English)
open.substack.com
May 6, 2025 at 11:35 AM
Increasing model size from 339M to 46B parameters enables models to grasp core 'protein grammar,' leading to higher-quality sequences that pass more rigorous filters.
Size matters! More parameters = better foundational understanding in protein design.
Size matters! More parameters = better foundational understanding in protein design.
Reposted by David
'AI Scientist' by FutureHouse (AI Literature search, with access to a large corpus of scientific texts, whereas most can only access abstracts)
It is free (for now)
Have you tried it? 👇
platform.futurehouse.org
It is free (for now)
Have you tried it? 👇
platform.futurehouse.org
FutureHouse Platform
AI Agents for Scientific Discovery
platform.futurehouse.org
May 1, 2025 at 11:38 PM
'AI Scientist' by FutureHouse (AI Literature search, with access to a large corpus of scientific texts, whereas most can only access abstracts)
It is free (for now)
Have you tried it? 👇
platform.futurehouse.org
It is free (for now)
Have you tried it? 👇
platform.futurehouse.org
Reposted by David
AGI AGI AGI.
Sam Altman and Dario Amodei see AGI coming as early as 2026.
Why so soon?
OpenAI’s definition = a technology that can perform most economically valuable tasks better than humans can.
But Demis Hassabis, CEO of Google DeepMind, is thinking bigger.
🧵👇
Sam Altman and Dario Amodei see AGI coming as early as 2026.
Why so soon?
OpenAI’s definition = a technology that can perform most economically valuable tasks better than humans can.
But Demis Hassabis, CEO of Google DeepMind, is thinking bigger.
🧵👇
April 18, 2025 at 6:55 PM
AGI AGI AGI.
Sam Altman and Dario Amodei see AGI coming as early as 2026.
Why so soon?
OpenAI’s definition = a technology that can perform most economically valuable tasks better than humans can.
But Demis Hassabis, CEO of Google DeepMind, is thinking bigger.
🧵👇
Sam Altman and Dario Amodei see AGI coming as early as 2026.
Why so soon?
OpenAI’s definition = a technology that can perform most economically valuable tasks better than humans can.
But Demis Hassabis, CEO of Google DeepMind, is thinking bigger.
🧵👇
Reposted by David
AI is no longer just a tool in the lab — it's becoming the lab.
At NVIDIA #GTC, the future of drug discovery was laid bare: computational, agentic, and lightning fast. Here's what you need to know 👇
At NVIDIA #GTC, the future of drug discovery was laid bare: computational, agentic, and lightning fast. Here's what you need to know 👇
April 18, 2025 at 10:30 AM
AI is no longer just a tool in the lab — it's becoming the lab.
At NVIDIA #GTC, the future of drug discovery was laid bare: computational, agentic, and lightning fast. Here's what you need to know 👇
At NVIDIA #GTC, the future of drug discovery was laid bare: computational, agentic, and lightning fast. Here's what you need to know 👇