AI Agent Architect: End-to-End LLM Solutions

Oppstar Berhad; Kajang; On-site; MYR 180,000 - 300,000; Full time

Job summary

Oppstar Berhad designs, builds, and optimizes domain-specific AI agents across semiconductor design, manufacturing, and internet products. You will own the end-to-end lifecycle of AI agents, from requirement discovery to architecture, tool integration, and continuous iteration based on user feedback.

Responsibilities include designing multi-step reasoning flows, integrating tools and APIs, and developing robust evaluation strategies for safety, reliability, and task success.

Qualifications

• Experience designing modular, scalable, and maintainable agent architectures for domain-specific applications.
• Experience with commercial and open-source LLMs.
• Proficiency in prompt engineering, chaining, and policy design.

Responsibilities

• Design multi-step reasoning flows, integrating external tools and APIs.
• Conduct prompt and policy optimization, and implement evaluation strategies for task success, safety, and reliability.
• Work with LLM platforms, RAG, and orchestration frameworks.
• Translate complex business and engineering use cases into practical AI agent solutions.
• Collaborate with domain experts to encode workflows and best practices into robust agentic systems.
• Connect agents to external tools, databases, and enterprise APIs.
• Implement multi-agent coordination and complex workflow automation.
• Design evaluation strategies: task success metrics, safety guardrails, reliability benchmarks, and user feedback loops.

Skills

Agent systems design

LLM platforms

Prompt engineering

RAG pipelines

Context fusion

Tool integration

Multi-agent coordination

A/B testing

Tools

LangChain

LlamaIndex

Semantic Kernel

Custom frameworks

Job description

Oppstar Berhad designs, builds, and optimizes domain-specific AI agents across semiconductor design, manufacturing, and internet products. You will own the end-to-end lifecycle of AI agents, from requirement discovery to architecture, tool integration, and continuous iteration based on user feedback.

Responsibilities include designing multi-step reasoning flows, integrating tools and APIs, and developing robust evaluation strategies for safety, reliability, and task success.