AI Computing Architecture Researcher

PERSOL SINGAPORE PTE. LTD.; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

PERSOL SINGAPORE PTE. LTD. is seeking an expert to design and optimize next-generation AI computing architectures and platforms for LLM, AIGC, and distributed AI workloads.

You will collaborate with researchers and engineers to improve performance across GPU, NPU, and heterogeneous environments. You will develop software components interfacing with hardware accelerators, improve performance, and contribute to research on agent capabilities and deployment.

Qualifications

• Master’s or PhD in Computer Science or Electrical Engineering.
• Strong experience in high-performance AI systems.
• Proficiency in C/C++, Python, and AI-DSL for low-level programming.
• Expertise in model optimization: quantization, pruning, sparsity.
• Experience with GPU/NPU programming and AI frameworks.
• Background in distributed systems and multi-agent coordination.
• Familiarity with containerization and cloud infrastructure.

Responsibilities

• Design and optimize AI computing architectures and platforms for LLMs, AIGC, distributed AI workloads, and intelligent agents.
• Develop software components interfacing with GPUs, NPUs, and AI chips.
• Improve performance and efficiency of distributed computing and GPU/NPU acceleration.
• Collaborate with AI researchers to enhance platform performance and resolve system-level issues.
• Design and implement tools for deployment, monitoring, scaling, and orchestration of AI workloads and agents.
• Integrate new AI models and algorithms ensuring scalability and fault tolerance.
• Conduct profiling and benchmarking to achieve high throughput and low latency.
• Contribute to research on agent capabilities including skill abstraction and deployment.

Skills

C/C++

Python

AI-DSL

CUDA

Triton

Distributed systems

Multi-agent coordination

PyTorch

TensorFlow

MXNet

Docker

Kubernetes

AWS

GCP

Azure

LLVM

MLIR

TVM

Agent skill deployment

Education

Master’s or PhD in Computer Science or Electrical Engineering

Tools

PyTorch

TensorFlow

CUDA

Docker

Kubernetes

SGLang

vLLM

HuggingFace

AWS

GCP

Azure

Job description

In this role, you will focus on designing and optimizing next-generation AI computing architectures and platforms for LLM, AIGC, distributed AI workloads, and intelligent agent systems. You will collaborate with researchers and engineers to improve the performance, scalability, and efficiency of AI infrastructure across GPU, NPU, and heterogeneous computing environments.

Key Responsibilities:

• Design and optimize AI computing architectures and platforms for LLM, AIGC, distributed AI, and intelligent agent workloads. Develop software components that interface with hardware accelerators such as GPUs, NPUs, and specialized AI chips.
• Improve performance and efficiency of distributed computing, parallel processing, and GPU/NPU acceleration.
• Collaborate with AI researchers to enhance platform performance for complex AI and agent-based applications and also to resolve system-level performance issues across architecture and platform components.
• Design and implement tools and frameworks for deployment, monitoring, scaling, and orchestration of AI workloads and agent systems.
• Integrate new AI models, agent controllers, and algorithms into the computing architecture and platform while ensuring scalability, fault tolerance, and efficiency.
• Conduct profiling, benchmarking, and performance optimization to achieve high throughput and low latency.
• Contribute to research on agent capabilities, including skill abstraction, composition, learning, and deployment.
• Stay updated on advancements in AI hardware, system architecture, and agent technologies, and continuously improve platform capabilities.

Required Qualifications:

• Master’s or PhD in Computer Science, Electrical Engineering, or related fields, with strong experience in high-performance AI systems or computing platforms.
• Proficiency in C/C++, Python, AI-DSL with a focus on low-level programming for high-performance systems.
• In-depth knowledge of AI model optimization techniques such as quantization, model graph pruning, and model parameter compression and sparsity algorithm.
• In-depth knowledge of parallel programming, distributed systems, and multi-agent coordination strategies.
• Experience with GPU/NPU programming (e.g., CUDA, Triton, cuTile or similar DSLs).
• Familiarity with AI/machine learning frameworks such as PyTorch, TensorFlow, or MXNet.
• Understanding of AI model deployment, orchestration, and optimization on large-scale platforms, including agent skill deployment and runtime management.
• Experience with containerization (Docker, Kubernetes), LLM deployment platforms (SGLang, vLLM, HuggingFace), and cloud infrastructure (AWS, GCP, Azure).
• Strong problem-solving skills and the ability to optimize software performance for both traditional AI and agent-based workloads.
• Experience with AI/deep learning frameworks and tools for performance profiling and optimization, including optimization in agent systems.
• Knowledge of low-level hardware optimization, including memory management and instruction-level tuning for CPU/GPU/NPU architectures.
• Familiarity with AI compiler and AI computing architecture, such as LLVM, MLIR, TVM.
• Background in designing and implementing high-performance distributed systems and storage solutions, including those supporting agent coordination and skill sharing.
• Strong understanding of networking, I/O, and data management techniques for AI and agent workloads.
• Exposure to multi-agent system design, including communication protocols, coordination mechanisms, skill composition frameworks techniques, harness engineering for AI workloads.