Research Engineer (FLASH)

NATIONAL UNIVERSITY OF SINGAPORE; Singapore; On-site; SGD 70,000 - 100,000; Full time

Job summary

The National University of Singapore is seeking full-time Research Fellows to develop hybrid physics-AI methods for weather applications. The role focuses on multimodal AI and foundation-model approaches using diverse data sources such as NWP outputs, satellite imagery, radar, and sensor networks.

You will build reproducible training pipelines, implement uncertainty quantification, and collaborate with weather-domain experts.

Qualifications

• PhD in Computer Science, Data Science, Engineering, Physics, or related.
• Strong Python and PyTorch; experience with multi-GPU/distributed training and performance optimization.
• Experience with real-world geospatial/sensor data (quality control, cleaning, visualization).
• Strong communication and collaboration skills.

Responsibilities

• Develop and benchmark multimodal AI / foundation-model approaches for spatiotemporal forecasting.
• Build reproducible AI training and evaluation pipelines, as well as uncertainty quantification strategies.
• Work at the intersection of physics and AI, with an emphasis on geospatial computational modelling.
• Collaborate with domain experts and operational stakeholders where relevant.
• Drive scientific breakthroughs and contribute to publications and cross-institutional collaborations.

Skills

Python

PyTorch

Multi-GPU training

Distributed training

Geospatial data handling

Communication skills

Education

PhD in Computer Science, Data Science, Engineering, Physics, or related

Job description

We regret that only shortlisted candidates will be notified.

Job Description

We are recruiting full-time Research Fellows to develop hybrid physics-AI methods for weather applications

Available data include:

• Numerical weather prediction (NWP) model outputs
• Weather satellite imagery
• Radar observations
• Lightning detection networks
• Surface sensor observations (e.g., rainfall and wind)

The successful candidates will:

• Develop and benchmark multimodal AI / foundation-model approaches for spatiotemporal forecasting.
• Build reproducible AI training and evaluation pipelines, as well as uncertainty quantification strategies.
• Work at the intersection of physics and AI, with an emphasis on geospatial computational modelling.
• Collaborate with domain experts and (where relevant) operational stakeholders.
• Drive scientific breakthroughs and contribute to publications and cross-institutional collaborations.

Qualifications

Required / strongly preferred

• PhD in Computer Science, Data Science, Engineering, Physics, or related.
• Strong Python and PyTorch; experience with multi-GPU/distributed training and performance optimization.
• Experience with real-world geospatial/sensor data (quality control, cleaning, visualization).
• Strong communication and collaboration skills.

Highly desirable

• Deep learning expertise: generative models, physics-aware learning, uncertainty modelling.
• Dense spatiotemporal prediction (e.g., video prediction, precipitation nowcasting).
• Atmospheric science / tropical meteorology background (a plus, not required).