Physical AI Heterogeneous System Design Engineer

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

Job summary

China Telecom Singapore Innovation Research Institute is seeking a pioneering Physical AI Heterogeneous System Design Engineer to redefine autonomous robotic manipulation through hardware acceleration, physical AI software models, and real‑time system control on heterogeneous GPU + FPGA testbeds.

You will design, develop, and deploy custom accelerators, integrate ROS 2 pipelines, and translate multimodal architectures into efficient C/C++ HLS specifications for robotic arms.

Qualifications

• Ph.D. or Master’s in Electronic Engineering, Computer Engineering, Robotics or related field.
• Proficiency in PyTorch / TensorFlow, ONNX model deployment workflows, and ROS / ROS 2 for robotic motion control.
• Demonstrated experience in AI model optimization, multimodal AI frameworks (ViT, CNNs, LLMs, VLMs) and simulation tools.
• 3+ years of hands‑on experience in High‑Level Synthesis (HLS) design.
• Proven record in hardware timing closure, resource optimization, and interfacing systems with robotic actuators.
• Experience in Physical AI and VLA models and hands‑on training/simulation for robotic arms.
• Strong programming skills in C, C++, and Python for embedded system drivers, hardware testbenches, and real‑time execution.

Responsibilities

• Develop, train, and optimize state‑of‑the‑art Physical AI models using PyTorch/TensorFlow and ONNX workflows.
• Design, implement, and optimize custom hardware acceleration cores using High‑Level Synthesis; achieve strict timing closure and low‑latency execution.
• Implement closed‑loop robotic arm motion control using ROS / ROS 2 and physics‑based simulation environments for data collection and evaluation.
• Lead end‑to‑end physical prototype demonstrations – integrating vision feeds, sensors, VLA results, and telemetry dashboards.
• Map microsecond real‑time motion control and sensor interfaces onto FPGA logic via PCIe and AXI4 interconnects.
• Translate high‑level multimodal neural architectures into C/C++ HLS specifications and ROS 2 middleware APIs for robotic control.

Skills

PyTorch / TensorFlow

ONNX

ROS / ROS 2

HLS

C

C++

Python

AI model optimization

multimodal AI frameworks

Education

Ph.D. or Master’s in Electronic Engineering, Computer Engineering, Robotics

Tools

MuJoCo

robosuite

Isaac Sim

Job description

We are looking for a pioneering Physical AI Heterogeneous System Design Engineer to redefine autonomous robotic manipulation through innovative hardware acceleration, physical AI software models, and real‑time system control. You will design, develop, and deploy custom hardware accelerators using High‑Level Synthesis (HLS) integrated with multimodal Vision‑Language‑Action (VLA) models and ROS 2 robotic arm control pipelines running on heterogeneous GPU + FPGA testbeds.

Key Responsibilities

• Develop, train, and optimize state‑of‑the‑art Physical AI models using PyTorch/TensorFlow and ONNX workflows.
• Design, implement, and optimize custom hardware acceleration cores using High‑Level Synthesis. Achieve strict timing closure and low‑latency execution.
• Implement closed‑loop robotic arm motion control using ROS / ROS 2 and physics‑based simulation environments for trajectory data collection and model evaluation.
• Lead the end‑to‑end physical prototype demonstration – integrating camera vision feeds, force sensor inputs, VLA real‑time inference results, and telemetry visualization dashboards.
• Map microsecond real‑time motion control and sensor interfaces onto FPGA logic via PCIe and AXI4 bus interconnects.
• Translate high‑level multimodal neural architectures into C/C++ HLS specifications and ROS 2 middleware APIs for robotic control.

Qualifications

• Ph.D. or Master’s in Electronic Engineering, Computer Engineering, Robotics, or a closely related discipline.
• Proficiency in PyTorch / TensorFlow, ONNX model deployment workflows, and ROS / ROS 2 for robotic motion control.
• Demonstrated experience in AI model optimization, multimodal AI frameworks (ViT, CNNs, LLMs, VLMs) and simulation tools (MuJoCo, robosuite, Isaac Sim).
• 3+ years of hands‑on experience in High‑Level Synthesis (HLS) design.
• Proven record in hardware timing closure, resource optimization, and interfacing systems with robotic actuators.
• Experience in Physical AI and VLA models and hands‑on training/simulation for robotic arms.
• Strong programming skills in C, C++, and Python for embedded system drivers, hardware testbenches, and real‑time execution.