Research Associate — AI for Science: Data Pipelines & HPC

NATIONAL UNIVERSITY OF SINGAPORE; Singapore; On-site; SGD 110,000 - 170,000; Full time

Job summary

National University of Singapore is seeking a research-engineer to design and build the Instrumentation Gym's foundational layers, enabling scalable HPC usage across labs. You will shape data pipelines, model interfaces, and baselines in collaboration with ML researchers and domain scientists.

The role involves developing LLM- and RAG-assisted tools for dataset navigation, and ensuring system scalability from single-GPU teaching to multi-node HPC environments.

Qualifications

• Degree in a Computational, Engineering, or Physical-Science discipline.
• A Master’s or Bachelor’s with equivalent research‑engineering experience.

Responsibilities

• Design and build the Instrumentation Gym substrate’s foundational layers for scalable HPC deployments.
• Create modular Python APIs to swap representation models and downstream components without rewriting pipelines.
• Develop baseline tokenization and cartography pipelines in collaboration with ML researchers and domain scientists.
• Build LLM- and retrieval-augmented interfaces for dataset navigation, workflow discovery, and user interaction.
• Ensure forward-compatibility to HPC with containerised, reproducible, scheduler-friendly deployments.
• Lead onboarding sessions and help instructors generate teaching material from Gym content.

Skills

Python

PyTorch

Data pipelines

GPU compute

Software maintainability

Communication with non-engineers

Scientific imaging data

HPC deployment

Slurm

Docker/Apptainer

Zarr/HDF5/OME-TIFF

Education

Master’s or Bachelor’s with equivalent research‑engineering experience

Tools

Slurm

Docker/Apptainer

HPC environments

HDF5

TIFF/OME‑TIFF

Zarr

Job description

National University of Singapore is seeking a research-engineer to design and build the Instrumentation Gym's foundational layers, enabling scalable HPC usage across labs. You will shape data pipelines, model interfaces, and baselines in collaboration with ML researchers and domain scientists.

The role involves developing LLM- and RAG-assisted tools for dataset navigation, and ensuring system scalability from single-GPU teaching to multi-node HPC environments.