Library

• Ever tried to assess an ingredient for safety … while wishing the data you need actually existed?
• Will safety assessors in cosmetics lose their jobs because of AI?
• How can you learn about Next Generation Risk Assessment (NGRA) for cosmetics safety?
• Why can great science succeed in pharma and still fail in personal care?
• The new Commission roadmap towards phasing out animal testing introduces "safe spaces".
• Europe’s overlooked deep tech is biology.
• I mapped 50 organisations shaping the beauty industry.
• Why B2B in cosmetics deserves more investor love.
• Not everyone in cosmetics openly shares the scientific research of their ingredient choices.
• What I’ve learned along the way - sharing seven lessons.
• How do you know founding a start-up is for you?
• One of 54 founders at the Claude: Life Sciences Hackathon.
• A few years ago, I knew almost nothing about federated learning.
• How to develop an entrepreneurial mindset as a scientist.
• A year ago, I made a (wild) bet on myself: to go all in on solving one of the toughest challenges in life sciences.
• Our work was recognised with the QSAR2025 oral presentation award.
• What will actually make AI work in life sciences?
• Here is a deep dive into key aspects of federated learning in life sciences.
• Is federated learning finally growing up in drug discovery?
• Are we finally seeing partnerships that put federated learning in life sciences to the test?
• What real-world problems is federated learning solving in drug discovery and development?
• How do we know which datasets in life sciences we still need but have not been created yet?
• How do you benchmark a federated model when even models trained on the same data can’t agree?
• The unspoken opportunity of federated learning in life sciences.
• 47 years, 1,099 opinions — and the big questions this raises for data, AI models, and regulatory trust.
• Cosmetic safety is not static — a timeline of SCCS itself and how policy and science shape each other.
• The ingredient categories the SCCS opinions covered — from hair dyes to nanomaterials, and what that means for safety assessment.
• Quantity vs. quality — why missing and inconsistent data remain the toughest barrier.
• Shifting definitions of “safe enough” — how SCCS classifications evolved, and why models must keep pace with regulatory language.
• What do SCCS opinions reveal about how we decide if a cosmetic ingredient is safe, and what does that mean for AI and risk assessment today?
• We will build better models from well-curated data and carefully designed experiments.
• How do you validate a predictive Bayesian model for skin sensitisation assessment, and what makes it scientifically reliable?
• If you're new to the SARA-ICE model for skin sensitisation assessment, this is a practical introduction to it.
• Three insights on what I learned from curating the largest open-source mechanistic data in skin sensitisation.

No posts match your search.