Senior AI Agent Architect - Multilingual Customer Service

SHOPEE SINGAPORE PRIVATE LIMITED; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

SHOPEE SINGAPORE PRIVATE LIMITED is seeking engineers to work on AI Agent architecture, Harness, and Loop Engineering for next-gen customer service solutions. You will design planning, tool use, memory, and multi-agent coordination, while reducing latency and improving model performance in real-world scenarios.

The role emphasizes Post-Training, RL, and RLHF/RLAIF methodologies, plus hands-on work with large language models and transformer architectures within a dynamic ecommerce data and

Qualifications

• Master’s degree or higher in CS/AI or related field.
• 3+ years of Algorithm Engineering or ML experience.
• Strong CS and math foundations with data structures knowledge.
• Solid ML/DL fundamentals, with understanding of Transformer and LLMs.
• Experience in Post-Training or RL/Agentic RL is a plus.
• Proficient in Python and familiar with PyTorch/TensorFlow.

Responsibilities

• Design and iterate AI Agent architectures and Harness components for customer service scenarios.
• Improve system efficiency with optimization of latency and cost.
• Develop self-evolution loops including data generation and evaluation.
• Explore RLHF/RLAIF in dialogue policies and agent decision points.
• Keep up with latest research and apply to practical workflows.

Skills

Python

PyTorch

Transformer

LLM architectures

Machine Learning

Algorithm Engineering

Reinforcement Learning

Context engineering

Education

Master's degree in Computer Science or related field

Tools

DeepSpeed

Megatron

TensorFlow

Job description

SHOPEE SINGAPORE PRIVATE LIMITED is seeking engineers to work on AI Agent architecture, Harness, and Loop Engineering for next-gen customer service solutions. You will design planning, tool use, memory, and multi-agent coordination, while reducing latency and improving model performance in real-world scenarios.

The role emphasizes Post-Training, RL, and RLHF/RLAIF methodologies, plus hands-on work with large language models and transformer architectures within a dynamic ecommerce data and