VP Research (Foundation Models)

Visium combines strategy, frontier AI, and deep regulated-industry expertise to build and deploy AI-native systems that reinvent how businesses operate - unlocking the value of AI at scale.

We're not a strategy firm that ships slide decks and walks away. Not a tech consultancy that builds what won't get adopted. Not a product company chasing problems it doesn't understand. We're all three at once: strategy, deep technical and integration expertise, and product, working as one. That combination is our flywheel. Consulting keeps us close to the hardest real-world problems. Products let us solve them at scale. Every engagement sharpens our products; every deployment makes the next engagement faster, deeper, and more certain.

We're obsessed with outcomes. While 95% of enterprise AI pilots never reach production, Visium is built to be the 5% delivering productised AI that compounds in value, not point solutions that deliver marginal gains. We are AI-native, not AI-adjacent. We don't bolt AI onto broken workflows; we redesign them with AI at the core.

Role

As a VP of Research - Foundation Models (Distinguished Scientist / Research Lead), you will own:

• Pre-training strategy and execution for biologics foundation models built from scratch across modalities (architecture, data, representation, scaling, evaluation).
• Continued pre-training and adaptation of vertical LLMs for regulated industries.
• The full training stack in practice: data pipelines, distributed training, scaling laws, evaluation harnesses, checkpointing, failure recovery.
• Research direction, standards and review for a small founding team.
• Translating frontier research into models that hold up in regulated, real-world use.

You’ll be a hands-on Research Lead / Distinguished Scientist role - not a manager of managers - owning architectures, training runs, and technical direction alongside a founding team of elite researchers. Freedom to publish, established compute resources, and direct collaboration with founders, offering real ownership and zero bureaucracy from day one.

To succeed in this role, you bring a combination of expertise, experience, and skills including:

• World-class pre-training experience (Meta / FAIR, OpenAI, Google DeepMind, Anthropic, Mistral, or equivalent/academic).
• Deep mastery of large-scale distributed training (parallelism strategies, scaling laws, optimization at scale, debugging instability).
• Demonstrated experience in data acquisition and preparation at scale.
• Strong publication or shipped-model track record.
• Biologics / scientific foundation model pre-training experience across modalities (e.g. protein, genomic, molecular, structural, multi-omics, or other specialized scientific domains). 
• PhD in a relevant field.
• Ability to code and stay hands-on in the details.
• Willingness to be based in Switzerland (Lausanne or Zurich), on-site.

By joining us as VP of Research, you'll benefit from

• Defined, in-place compute to run real pre-training.
• A founding research seat with genuine technical authority.
• Direct partnership with top-20 pharma companies to bring world-leading models to life.
• Direct line to the founding team and co-shaping the company strategy
• Ability to publish and open source research
• A competitive compensation package incl. Equity
• A flexible working culture of freedom & responsibility
• A position that enables you to have an impact on 1’000s of people and the future of the company
• An international, passionate team of demonstrated extraordinary achievements with a strong collaborative mindset and a drive to make a positive difference