Stefano Massaroli

@Massastrello

Co-founder and President @RadicalNumerics | Researcher @ RIKEN | AI ∩ numerics ∩ signal processing

Bunkyo-ku, Tokyo

radicalnumerics.ai

Joined July 2019

Pinned
Stefano Massaroli
@Massastrello
Jun 15
@RadicalNumerics has raised $50 million in Seed funding. Evolution is the longest optimization run in history: four billion years of search toward an objective no one chose. We are here to design what evolution didn’t — and the defenses to keep that power safe. Both are
Eric Nguyen
@exnx
Jun 15
Together with my co-founders Michael @MichaelPoli6, Stefano @Massastrello and Armin @athmsx, I am excited to announce @RadicalNumerics is emerging from stealth with a $50M seed round to build general biological intelligence. We’re also sharing an early preview of our new model
Stefano Massaroli
@Massastrello
Jun 17
For a single toxin there are more working disguises than atoms in the universe -- sequences that kill the same way but match nothing on record. You can't screen a list against infinity; bio-defense has to understand function, not compare strings. That's exactly what Omnii does.
Radical Numerics
@RadicalNumerics
Jun 17
1/ Leaders of OpenAI, Anthropic, & GDM signed a letter urging Congress to mandate DNA synthesis screening. But chatbots & agents can't read DNA. So we built Omnii to defend against natural & AI-designed pathogens. Blog: radicalnumerics.ai/blog/omnii-def… Letter: screendna.org
GIF
Stefano Massaroli
@Massastrello
Mar 19
Really proud of the progress we’re making at @RadicalNumerics. AI has made incredible strides in the world of bits. But science lives in the world of atoms — messy, complex, and grounded in physical reality. Bridging that gap requires a fundamentally different approach to
Michael Poli
@MichaelPoli6
Mar 18
We're growing rapidly at @RadicalNumerics and scaling our core teams. Join us in building the next generation of scientific world models. We're hiring across a few roles, each with significant ownership and cross-functional scope: - Member of Technical Staff, Post-Training -
Stefano Massaroli
@Massastrello
Jan 12
NVIDIA's NVFP4 enables stable 4-bit pretraining on Blackwell—matching higher-precision quality with far less memory. This unlocks massive scaling for scientific AI: complex distributions and huge datasets become tractable at frontier scales. At Radical Numerics, we're building
Radical Numerics
@RadicalNumerics
Jan 12
Scaling scientific world models requires co-designing architectures, training objectives, and numerics. Today, we share the first posts in our series on low-precision pretraining, starting with NVIDIA's NVFP4 recipe for stable 4-bit training. Part 1: radicalnumerics.ai/blog/nvfp4-par… Part
Stefano Massaroli
@Massastrello
Aug 25, 2025
I co-founded @RadicalNumerics with @MichaelPoli6, @ai_with_brains, @exnx & a stellar team of researchers and engineers. AI is changing everything—except how it changes itself. We’re building the infrastructure to automate the evolution of intelligence.