Research Scientist (AI Model Optimization), IPV, ARTC

A*STAR RESEARCH ENTITIES; Singapore; On-site; SGD 150,000 - 210,000; Full time

Job summary

A*STAR RESEARCH ENTITIES is seeking a PhD-level researcher to lead optimization work for high-performance AI. You will explore QAT, pruning, distillation, and NAS to reduce latency while designing scalable AI deployment architectures for high-throughput camera and sensor data.

You will co-design hardware/software for deployment on Jetson/TensorRT/FPGAs and build asynchronous data pipelines, with rigorous benchmarking. Expect collaboration with SI and impact across manufacturing environments.

Qualifications

• PhD in Computer Engineering, Computer Science, Electrical Engineering or related field with HP AI focus.
• Minimum 3 years of experience in relevant tech and research roles.
• Deep understanding of AI inference engines and model compression techniques.
• Proficient in C++ and Python for high-performance implementation.
• Hands-on parallel computing with CUDA/OpenCL and mixed-precision deployment.

Responsibilities

• Lead research into optimization techniques (QAT, pruning, distillation, NAS) to minimize latency.
• Design scalable AI deployment architectures for multi-camera/high-throughput data streams.
• Co-design hardware/software for deployment targets (Jetson, TensorRT, FPGAs, accelerators).
• Develop asynchronous data pipelines from image acquisition to decisions.
• Establish performance benchmarks for latency and memory across environments.
• Collaborate with System Integrator to integrate models into factory software stack.

Skills

AI inference engines

Model compression

C++

Python

Parallel computing

Mixed-precision training

End-to-end AI systems

Education

Ph.D. in Computer Engineering/CS/EE or related field (High-Performance AI)

Tools

TensorRT

ONNX Runtime

OpenVINO

CUDA

OpenCL

Job description

Responsibilities

• Lead research into state-of-the-art optimization techniques, including Quantization-Aware Training (QAT), Pruning, Knowledge Distillation, and Neural Architecture Search (NAS) to minimize latency.
• Design and implement scalable AI deployment architectures that can handle high-throughput data streams from multiple high-resolution cameras and process sensors simultaneously.
• Conduct hardware-software co-design to optimize models for specific deployment targets (e.g., NVIDIA Jetson, TensorRT, FPGAs, or specialized AI accelerators).
• Develop and manage asynchronous data pipelines that ensure zero-bottleneck performance from image acquisition to "final sentencing" decisions.
• Establish rigorous performance profiling benchmarks to track model latency and memory footprint across various manufacturing environments.
• Work with the System Integrator (SI) to ensure that optimized models are seamlessly integrated into the factory-level software stack.

JOB REQUIREMENTS

• Ph.D. in Computer Engineering, Computer Science, Electrical Engineering, or a related field with a focus on High-Performance AI.
• At least 3 years of experience
• Deep understanding of AI Inference Engines (e.g., TensorRT, ONNX Runtime, OpenVINO).
• Mastery of Model Compression techniques (Pruning, Quantization, Distillation).
• Expertise in C++ and Python for high-performance implementation.
• Hands-on experience with Parallel Computing (CUDA, OpenCL).
• Familiarity with Mixed-Precision Training and FP16/INT8 deployment.
• Proven ability to architect end-to-end AI systems that balance the trade-off between throughput, latency, and model precision.

The above eligibility criteria are not exhaustive. A*STAR may include additional selection criteria based on its prevailing recruitment policies. These policies may be amended from time to time without notice. We regret that only shortlisted candidates will be notified.