Co-Founder, CTO, Physical AI Process Industries

About Marble

Marble is a climate tech venture studio. We partner with scientists, engineers, and operators to create companies solving hard climate problems in the world's largest industries.

Our first 14 companies are building transformative products across energy, industry, agriculture, and climate resilience. 100% have been successful in raising follow-on to date. We are looking to partner with exceptional individuals to create our next venture.

The opportunity below is a founding role in our next company creation, which we think will be absolutely massive!

Process industries - food & beverage, chemicals, pharmaceuticals, metals, and materials - form the backbone of the global economy. In the EU alone, they generate €5T in annual revenue and consume 75-80% of all industrial energy. But inside these plants, critical decisions are still driven by conservative operating conditions, fragmented data, and operator intuition.

As a result, 15-25% of plant revenue (~€1T annually) is lost to inefficiencies from over-cleaning, undetected fouling, suboptimal process conditions, and siloed optimisation. We want to support industrial competitiveness by focusing on addressing these process inefficiencies.

These problems aren’t caused by a lack of data. Most plants produce continuous data from thousands of sensors, but that data sits locked in silos. Integration is slow, contextualisation is poor, analysis models are bespoke, and replication across sites takes months. We are building a new platform using AI and innovative modelling to solve these challenges, creating better decisions based on existing data and enabling it to scale across plants, faster.

We are building the physical AI for process plants: a system that combines physics-based models, live sensor data, and machine learning to make equipment behaviour observable and provide real-time recommendations across operations.

We have identified a key entry point that has previously been overlooked. Where customer pain is high, and deployment is fast. Solving it delivers immediate savings across energy, water, chemicals, and downtime. And we’re already signing paid pilots. This early use case allows us to build the foundations of our platform.

From there, our approach extends to adjacent unit operations within the plant and across industries, with the long-term vision of a unified layer that optimises process plant operations in real time. At scale, we address a €300B market, reduce industrial energy spend by 15%, and tackle 4% of global CO₂ emissions.

We are looking for a CTO and Co-Founder with world-class ML engineering experience who wants to apply their talent to improving the physical world. You will take full ownership of the technical direction and execution, from early pilots to full-scale deployment. You will shape the platform models and architecture (data curation & contextualisation, physics models, ML pipelines, operator-facing apps), focused on building a robust, replicable product.

As we scale, you will build the engineering team and set the bar for technical excellence.

You will join our current Founder in Residence and CEO: a PhD in process engineering from ETH Zurich, with experience working on the factory floor across pulp & paper, waste-to-energy, cement, and food production. He has developed hybrid physics-ML models and optimised production lines. He brings deep knowledge of both the physics and the customer, and an obsession with making the industry better.

We are looking for an entrepreneurial Applied AI/ML Engineer who can turn ambiguous problems into a technical vision and operational reality. You have a driving motivation to build the future of industrial process plant operations.

Must have:

• Deep ML expertise across at least two of physics-informed models, time‑series, reinforcement learning, or agentic systems; and the judgement to choose the right tool not the fanciest.
• Comfortable turning dirty data into reliable training sets through sensor data curation, plant‑graph contextualisation, and integration with semi‑structured records.
• AI‑native by default,…