Remote World-Model AI Researcher — Built Environment

Geolava; United States; Remote; USD 180,000 - 250,000; Full time

Benefits offered by this job

Medical, dental, vision insurance

Retirement plan

Global remote workforce

Job summary

Geolava is building spatial intelligence systems and is seeking an AI Researcher to own end-to-end research from framing questions to production-ready capabilities. You will develop world models using multi-temporal data and train large vision models to forecast asset changes over time.

The role emphasizes autonomous work on ambiguous problems in a startup context, with collaboration to ship capabilities for investors, lenders, and infrastructure owners.

Qualifications

• World-model research experience with academic or industry work.
• PhD in machine learning, computer vision, or closely related field.
• Fluency with deep learning for large vision models; diagnosing training runs.
• Empirical rigor with proper baselines and data splits.
• Ability to work autonomously on ambiguous problems in a small team.
• Clear written and spoken communication for technical and non-technical audiences.

Responsibilities

• Design, train, and evaluate world models of assets changing over time.
• Develop representation learning over multi-temporal imagery and data.
• Build rigorous evaluation with baselines, temporal splits, calibration, and leakage checks.
• Run experiments on cloud GPUs with attention to throughput and cost, produce writeups.
• Collaborate with AI and engineering teams to ship research as platform capabilities.
• Stay current with world-model, self-supervised, and generative modeling literature.

Skills

World-model research experience

Deep learning fluency

Empirical rigor

Ownership

Clear communication

Education

PhD in ML/CV or related field

Tools

PyTorch

Vision transformers

Diffusion models

Python

Cloud GPUs

Job description

Geolava is building spatial intelligence systems and is seeking an AI Researcher to own end-to-end research from framing questions to production-ready capabilities. You will develop world models using multi-temporal data and train large vision models to forecast asset changes over time.

The role emphasizes autonomous work on ambiguous problems in a startup context, with collaboration to ship capabilities for investors, lenders, and infrastructure owners.