Applied Machine Learning Researcher - Multimodal Foundation Models

Location: San Francisco Bay Area, California

The Opportunity

We're working with a venture-backed AI company developing a new generation of foundation models designed to understand and reason about the physical world.

While much of today's AI research focuses on language and vision, this team is exploring how models can learn from a much broader range of information, including sensor readings, time-series data, audio and other signals generated by real-world systems.

The aim is to develop models capable of interpreting complex physical behaviours, understanding how systems change over time and generating insights that can be applied to real industrial challenges.

With a growing research and engineering team, our client is looking for an Applied Machine Learning Researcher who can help shape these models from early experimentation through to real-world deployment.

This is an opportunity to work on challenging foundation model research, with the freedom to explore new approaches and see your work translate into practical applications.

The Role

You'll take ownership of research and model development across multimodal learning, working at the intersection of foundation models, time-series modelling and physical-world intelligence.

The position combines scientific research with hands-on engineering. You'll be involved in defining research problems, designing experiments, developing and training models, evaluating their performance and helping bring successful approaches into production.

You'll also have the opportunity to influence the team's wider research direction and contribute to work with publication potential.

What You'll Be Doing

• Research, develop and improve multimodal foundation models combining sensor and time-series data with language, vision, audio and other modalities.
• Take ownership of end-to-end ML research, from problem formulation and data preparation through to model training, experimentation and evaluation.
• Explore novel model architectures and training approaches that improve understanding of complex physical systems.
• Develop methods that enable models to reason across long sequences of data, multiple information sources and changing real-world conditions.
• Design rigorous experiments, evaluate model performance and identify opportunities to improve generalisation and reliability.
• Work with engineering teams to scale successful research into production models supporting industrial and real-world applications.
• Help shape future research priorities and contribute to technically meaningful work with potential for publication.

What We're Looking For

We're interested in researchers and applied ML engineers who combine strong scientific thinking with the ability to build and develop models independently.

Ideally, you'll bring:

• Strong experience in machine learning research, with the ability to formulate hypotheses, design experiments and independently investigate new approaches.
• Hands-on experience across the ML development lifecycle, including data preparation, model development, training and evaluation.
• A solid understanding of modern deep learning architectures, training methods and model evaluation.
• Experience taking machine learning models beyond experimentation and towards deployment or production.
• Strong technical judgement, experimental discipline and the ability to work through open-ended research challenges.
• The confidence to take ownership of complex modelling problems and contribute to technical decisions.

Particularly relevant experience would include:

• Multimodal learning or foundation model development and adaptation.
• Time-series modelling, sensor data, signal processing or other non-text data modalities.
• Research involving transformers, long-context modelling or cross-modal representation learning.
• Experience applying advanced ML models to physical systems or real-world datasets.

Previous experience across every modality isn't essential.

We're particularly interested in people who can demonstrate strong independent research, sound experimental judgement and ownership of their modelling work.

Why Consider This Opportunity?

This is a chance to work on an emerging area of machine learning where many of the core research challenges are still being explored.

You'll be joining a specialist team building foundation models that can interpret information beyond conventional text and images, with applications across complex physical and industrial environments.

The role offers considerable technical ownership, exposure to challenging multimodal research problems and the opportunity to see your ideas progress from initial experiments into models used in real-world settings.

For someone who enjoys the research process but also wants to understand how their work performs outside the lab, it offers a compelling combination of scientific depth and practical engineering impact.