AI Research Scientist / Engineer Generative World Models

present of AI; København; Hybrid; DKK 550,000 - 950,000; Full time

Job summary

present of AI seeks an AI Research Scientist / Engineer to advance generative world models using behavioral data from care settings. You will focus on training and scaling large sequence models and turning representations into practical predictions like short-term fall risk.

You will develop PyTorch-based pipelines, design experiments, and ship code while addressing irregular time data, long-context models, and distributed training challenges in a research-driven environment.

Qualifications

• Experience with PyTorch and solid grounding in probability and statistics.
• Ability to design experiments, read results honestly and ship working code.
• Experience handling irregular time data, long context models, or distributed training.

Responsibilities

• Train and scale large sequence models on behavioral data from care settings.
• Turn learned representations into useful predictions such as short term fall risk.
• Design and run experiments, validate results, and deliver production-ready code.

Skills

PyTorch

Statistics

Experiment design

Distributed training

Time-series data

Job description

AI Research Scientist / Engineer Generative World Models

The role centers on training and scaling large sequence models on behavioral data from care settings, then turning the learned representations into useful predictions such as short term fall risk. A strong candidate has hands-on PyTorch experience, solid grounding in probability and statistics, and the independence to design experiments, read results honestly and ship working code. Experience with irregular time data, long context models or distributed training is a clear plus.

How it fits the AI era

The work sits right at the core of the AI era, since the same ideas behind language models are being applied to human behavior over time. AI handles the pattern learning at scale, while people still choose which questions matter, judge whether a prediction is trustworthy, and decide how a forecast should shape care. Good research judgment and care for real patients stay human responsibilities.

How it builds AI experience

Someone in this role would gain deep experience pretraining a foundation model on a proprietary stream of real world camera based behavior data, including tokenization, scaling experiments and taking a model into live use. To build toward it, practice training small sequence models on event or time series data, try continuous time methods, and write up what the experiments show.

Worth showing

• Training large sequence models
• PyTorch training pipelines
• Time-to-event or point processes
• Distributed GPU training experience

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About Teton

Per the company, Teton provides an AI and computer vision platform that monitors patient and resident rooms to detect activity and risk in real time.