Research Scientist, Foundation Model (World Model & Site Memory)

Startupvalleys; Zürich; Hybrid; CHF 140,000 - 210,000; Full time

Benefits offered by this job

Equity

Salary package

Ownership of projects

International team

Job summary

Laelaps AI in Zürich, Switzerland is seeking a Research Scientist to advance the foundation model of the physical world. You will own questions around site memory, anomaly detection, and world-models, and you will ship results directly to production alongside perception and model engineers.

You will work on probabilistic, self-supervised spatio-temporal modelling, and apply these ideas to real security sites with live data from cameras and sensors.

Qualifications

• PhD in machine learning, computer vision, robotics or a related field with first-author publications at top venues.
• Experience with world models, video or latent prediction, self-supervised representation learning, or probabilistic modelling of spatio-temporal data.
• Experience with anomaly or novelty detection where positives are rare.
• Strong PyTorch and the engineering to run your own experiments at scale.

Responsibilities

• Own the research question at the center of LEA, the foundation model of the physical world.
• Develop site memory as a probabilistic model of people, vehicles and robots by hour and weather.
• Explore anomaly as-surprise and evaluate when real incidents are rare.
• Train world-model weights on multi-sensor field sequences and transfer across sites.
• Shape the research direction, design experiments and decide what is true before shipping.

Skills

PhD in ML

World models

Self-supervised learning

Anomaly detection

PyTorch

Education

PhD in ML/CS/Robotics

Tools

PyTorch

CUDA

Job description

Research Scientist, Foundation Model (World Model & Site Memory) at Laelaps AI · Startup Valleysstartupvalleys.comMapCompaniesJobsValleysGet featuredManage companySign inResearch Scientist, Foundation Model (World Model & Site Memory)at Laelaps AIApplyAll rolesNext rolePosted 29 Oct 2026Laelaps AI Zürich, Switzerland · MapApply# Research Scientist, Foundation Model (World Model & Site Memory)ResearchOriginal posting — from Laelaps AIOpen original#### Our MissionAt Laelaps AI, we believe robotics is entering a transformative decade, much like the arrival of the internet. Advances in AI, cloud computing, and hardware are reshaping what autonomous systems can do. Our mission is to build the intelligent software that powers physical security in the real world - enabling robots and sensors to handle dangerous and critical tasks that humans shouldn't have to. By engineering the orchestration layer for intelligent security, we aim to create a world that is safer, more secure, and more resilient.We're a strong founding team based in Zurich, backed by visionary investors and advisors. We are engineering the future of security today!#### The RoleYou will own the research question at the centre of LEA, our foundation model of the physical world: can a model learn what normally happens on a site, by place, hour and weather, so that an anomaly is simply what it did not expect? The first version is a probabilistic memory of each site. The next is a world model across every site that predicts the next state in latent space, starting from open world-model weights and growing on field data nobody else holds.Our robots and cameras run on live security sites around the clock, and every event is kept with its clip, the model's verdict and the operator's outcome. That record is your dataset. You will work alongside our perception and model engineers, and your results go into production, not only into papers.#### What You'll Work On* **Site memory:** a learned, probabilistic model of each site: which people, vehicles and robots are where, at what hour, in what weather, and how the scene usually unfolds.* **Anomaly as surprise:** replace rule-based alerting with measured surprise, and work out how to evaluate it when real incidents are rare.* **World model:** post-train open world-model weights on aligned multi-sensor field sequences, JEPA-style, predicting in latent space rather than pixels.* **Transfer:** from one site's memory to a model of all of them, and representations that carry across sites and domains.* **Research direction:** choose the questions, design the experiments, and decide what is true before it ships.#### Who We're Looking ForA research scientist who has led work on world models, self-supervised learning or probabilistic spatio-temporal modelling, and who wants to see it run on real sites. You are rigorous about what a result does and does not show, and pragmatic about getting it into a product.#### Your Background* PhD in machine learning, computer vision, robotics or a related field, with first-author publications at top venues.* Research experience in world models, video or latent prediction, self-supervised representation learning, or probabilistic modelling of spatio-temporal data.* Experience with anomaly or novelty detection where positives are rare.* Strong PyTorch and the engineering to run your own experiments at scale.#### Nice to Have* Scene graphs, multi-object tracking or trajectory forecasting.* Bayesian or probabilistic deep learning.* Robotics or embodied AI.* Simulation for data generation.#### What We Offer* **Ownership:** you are able to ship products and deliver project end-to-end.* **Mission:** autonomous security that keeps people and critical sites safe, including in defence.* **Career path:** a ground-floor seat with real runway. Prove your value and you will not have barriers to grow.* **Team:** work directly with PhD-level co-founders in AI, Robotics, and Physics, alongside a strong (and fun) founding team.* **Compensation:** Competitive equity/salary package* **Culture:** International founding team that is serious about building but does not take itself too seriously.This provider only shows the posting on its own website.