Machine Learning Applied Researcher

Archetype AI; San Mateo (CA); On-site; USD 120,000 - 180,000; Full time

Job summary

Archetype AI in San Mateo is seeking an experienced ML researcher to build multimodal foundation models integrating real-world signals. This role requires a self-directed candidate capable of leading research from problem formulation through deployment. Key responsibilities include model development, evaluation, and scaling for industrial applications. Ideal candidates have end-to-end modeling experience and comfort with ambiguous problem spaces. Join us to impact the next generation of AI grounded in physical reality and complex systems.

Qualifications

• Ability to operate in ambiguous problem spaces.
• Experience with productionization or deployment of ML models.
• Strong experimental rigor in scientific work.

Responsibilities

• Build and improve multimodal foundation models using various data types.
• Drive research and modeling efforts from definition to evaluation.
• Scale research experiments to production models.

Skills

Self-directed

Strong scientific work

End-to-end modeling

Multimodal experience

Technical judgment

Job description

About Archetype AI

Archetype AI is developing the world’s first AI platform to bring AI into the real world. Formed by an exceptionally high-caliber team from Google, we are building a foundation model for the physical world—a real-time multimodal LLM that transforms real-world data into valuable insights. It will help people in their real lives, not just online, by understanding the real-time physical environment and everything that happens in it.

Supported by deep-technology venture funds in Silicon Valley, we are at Series A and rapidly advancing toward the next stage. Our headquarters are in San Mateo, California, with team members across the US and Europe. We are actively growing and welcome exceptional candidates who are excited to work on cutting-edge physical AI. If a role does not exactly match your experience, you may contact us directly with your resume via jobsarchetypeaiio.

About the Job

We are building a new class of multimodal foundation models for the physical world, combining time-series/sensor data, language, vision, audio, and other real-world signals into unified models that can understand complex systems, reason over long horizons, and support real-world tasks in industrial and physical environments.

We are looking for an experienced, researcher-oriented ML candidate to help build these systems end to end—from problem formulation and experimental design to model development, evaluation, and deployment.

This role is intended for someone who is highly self-directed, can independently perform strong scientific work, and is excited to work on multimodal intelligence grounded in physical signals.

What You’ll Work On

• Build and improve multimodal foundation models that incorporate time-series/sensor data alongside language, vision, audio, and related modalities.
• Drive research and modeling efforts from problem definition through experimentation and evaluation.
• Own modeling work across data, modeling, and evaluation.
• Advance model architectures and training strategies for physical-world understanding and long-context reasoning.
• Drive and scale research experiments and modeling advances to production models that power diverse use cases in complex industrial scenarios.
• Contribute to research directions with potential for publication.

What We’re Looking For

• Self-directed and comfortable operating in ambiguous problem spaces.
• Able to independently perform strong scientific work, including forming hypotheses, designing experiments, and drawing sound conclusions.
• Experience with end-to-end modeling, including data, modeling, and evaluation.
• Experience with productionization or deployment of ML models.
• Multimodal experience preferred.
• Strong technical judgment and experimental rigor.

Why this role is interesting

• Many important real-world systems cannot be understood from text or vision alone; they depend on signals that evolve over time—sensors, operating conditions, environment, and interactions across subsystems. The next generation of useful foundation models will need to integrate these sources and reason over them in a unified way.
• You will have a unique opportunity to help shape a new generation of multimodal foundation models grounded in physical signals and real-world dynamics. The role offers a rare combination of deep research challenges, practical deployment impact, and the chance to contribute to a fast-emerging area of AI.