Sigma Nova

Our Vision: Interdisciplinary AI for Complex Domains

We believe the future of AI isn't just about scaling compute or scraping more internet data. The most impactful breakthroughs will come from domains where:

• Data is complex and multimodal (think brain signals, molecular structures, industry machinery sensor data, astronomical observations)

• Data is scarce or hard to collect (highly regulated industries, expensive instrumentation)

• Impact is enormous (healthcare, climate, materials science, robotics)

These domains share common challenges from research, data, and technical standpoints. We're building reusable blueprints and foundational building blocks that can transfer across scientific fields—what we learn from brain modelling can inform physics, chemistry, or beyond.

Starting with the Brain

Our first proving ground is AI for neuroscience. We're encouraged by early partnerships with research labs and our team's performance in the EEG NeurIPS Challenge December 2025), where we achieved beyond state-of-the-art results.

The challenge we're tackling is urgent: Brain illness and neurodegenerative disorders represent one of the most threatening health crises of our era. Applying frontier AI to complex, multimodal brain datasets could unlock transformative clinical applications.

But we're also investing in theoretical foundations. Our team is publishing in top-tier ML conferences, ex:

• Paper 1: arxiv.org/abs/2511.17339

• Paper 2: arxiv.org/abs/2510.04602

This dual focus—rigorous science and applied impact—defines our approach.

Our Business Model: Research-First, Revenue-Informed

We are, above all else, a science based company. Not a product factory. Not a consultancy in disguise. Research is our reason to exist.

But to sustain this mission long-term, we need financial viability. That doesn't mean compromising our science—it means being strategic about how we capture value from the impact of our research.

Three Revenue Streams

1. IP Valorization: Selling the Components

Our research produces valuable building blocks that can be licensed or sold independently:

• Modular architectures and model components

• Training pipelines and tooling

• Proprietary techniques for data processing and fine-tuning

• Foundation Model blueprint

We monetize through licensing agreements, APIs, or partnerships. Think of it as selling the LEGO blocks, not the finished castle.

2. Foundation Models as Products

When a foundation model reaches maturity and has clear market potential, it can become a standalone product—either within Sigma Nova or as a spin-off/spin-out company. These aren't side projects; they're potential businesses built on our core research.

Our brain models, for instance, could eventually power diagnostic tools, drug discovery platforms, or clinical decision support systems in Life Science and Healthcare.

3. Research Consulting Services

We work directly with large French enterprises that have their own research teams. This isn't generic AI consulting—it's deep collaboration on models and architectures, helping corporate research labs tackle problems they can't solve alone.

This keeps us close to real-world challenges, funds ongoing science, and builds strategic relationships—all while protecting our research independence.

We need people who:

• Are excited by fundamental challenges, not just quick wins

• Want to do rigorous science that also matters in the real world

• Thrive in small, interdisciplinary teams where your work has immediate impact

• Understand that breakthrough research takes time, iteration, and patience

We're not looking for people chasing rapid equity rewards or galactic hypergrowth. We're ambitious, but we're realistic about the enormous challenge ahead.

Right now we're 8 people. By the end of 2026, we aim to be about 40 (new offices coming). We're hiring brilliant Frontier AI scientists and, increasingly, exceptional research engineers who can scale our ML practices—PyTorch-heavy, running on French infrastructure Scaleway and GENCI GPUs.

But if you want to work on genuinely hard problems at the intersection of AI and science, surrounded by brilliant people and backed by partners who understand the long game, let's talk.