Portrait of Quentin de Laroussilhe

Quentin de Laroussilhe

ML Research & Engineering

I build large-scale machine learning systems that make models more adaptive, reusable, and efficient. At DeepMind, I work on continual learning and modular training, developing methods for systems that can accumulate knowledge and generalize across tasks.

Before DeepMind, I co-founded and led the engineering team behind automated ML workflows across Ads, YouTube, and Gmail, after several years in Google Brain working on neural architecture search, transfer learning, and the early Google Assistant.

Connect on LinkedIn View Google Scholar
ANGEL INVESTING

Alongside my career in machine learning, I invest independently in early-stage deep tech ventures.

Portfolio Thesis

I back founders tackling problems with long-term scientific moats, the kind I’d be excited to work on myself.

I look for clarity, technical depth, and defensible innovations with the potential to create large-scale societal or geopolitical impact.

Selected Investments

Zama

Homomorphic encryption infrastructure
Seed Exited

Enables computation on encrypted data, creating the foundation for privacy-preserving AI and digital sovereignty.

Brink Tx

Programmable recombinase gene editing
Seed

Evolves recombinases for in-vivo precise gene edits, unlocking scalable and safe genetic therapies.

Blue Morpho

AI infrastructure for structured knowledge
Seed

Transforms unstructured text into structured, queryable knowledge, essential for reliable reasoning and automation.

Vortex Aerotec

Hydrogen-powered VTOL logistics
Pre-seed

Designs dual-use long-range autonomous aircraft for industrial and defense logistics, ensuring logistical continuity.

Get in touch

I read every deck myself. I’m initially looking for alignment with my thesis. If it resonates, I’ll get in touch to explore it further.

Thanks, your deck was sent!

If you’re a fund or co-investor interested in connecting, feel free to reach out on LinkedIn.