Physical AI Research Scientist

China Telecom Singapore Innovation Research Institute; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

China Telecom Singapore Innovation and Research Institute invites applications for a Physical AI (Edge Intelligence) Research Scientist to push real-world capabilities of AI models on edge devices. You will bridge frontier AI research with hardware deployment, optimizing neural models and edge inference pipelines.

You will collaborate with hardware acceleration engineers, focusing on high-impact AI model research, algorithmic optimization, edge validation, and publishing in top conferences.

Qualifications

• PhD in CS/AI or related with strong publication record.
• Strong DL, multimodal AI, or foundation models background.
• Proficiency in PyTorch and model export/deploy workflows (ONNX/TensorRT).

Responsibilities

• Lead research, adaptation of multimodal foundation models for edge physical systems.
• Implement model compression, quantisation, and graph optimisation for low-latency edge execution.
• Deploy and benchmark AI models on physical edge testbeds with vision feeds and multimodal inputs.
• Utilise simulation environments for synthetic data generation and benchmark evaluation.
• Publish high-impact research in premier AI/ML conferences and formulate technical patent disclosures.

Skills

Deep Learning

Multimodal AI

Foundation Models

PyTorch

Python

C/C++

Edge AI

Publications

Education

PhD (or research Master's with publications)

Tools

ONNX

TensorRT

Job description

Role Overview

China Telecom Singapore Innovation and Research Institute (CTSIRI) is seeking a Physical AI (Edge Intelligence) Research Scientist to advance the frontier of real-world physical intelligence. In this role, you will lead the research, adaptation, and deployment of foundation AI models for low-latency execution on edge physical systems.

You will bridge the gap between frontier AI research and practical hardware implementation, developing robust neural models and optimising edge inference pipelines. You will collaborate closely with in-house hardware acceleration engineers, allowing you to focus on high-impact AI model research, algorithmic optimisation, edge validation and top-tier scientific publications.

Key Responsibilities

• Spearhead research, fine-tuning, and adaptation of multimodal foundation models and action-generating policies for edge physical systems.
• Implement model compression, quantisation, and graph optimisations to enable low-latency execution on edge hardware.
• Deploy and benchmark AI models on physical edge testbeds, integrating vision feeds and multi-modal sensory inputs.
• Utilise simulation environments for synthetic data generation, pre-training, domain adaptation, and benchmark evaluation.
• Publish high-impact research in premier AI/ML conferences and formulate technical patent disclosures.

Qualifications

• Ph.D. (or research Master's with a strong publication record) in Computer Science, Artificial Intelligence, Computer Vision, Electrical Engineering, or a related quantitative discipline.
• Strong research foundation in Deep Learning, Multimodal AI, or Foundation Models (e.g. Transformers or Multimodal LLMs/VLMs).
• Solid track record of publications in conferences or journals.
• Proficiency in PyTorch and familiarity with standard model export and deployment workflows (e.g. ONNX or TensorRT).
• Strong programming skills in Python, with working familiarity in C/C++.
• Demonstrated interest or experience in evaluating AI models on physical edge systems, hardware platforms, or testbeds.