Job Responsibilities
• Design and build a reusable multi-agent AI system that supports multiple projects and use cases
• Define agent roles, responsibilities, and interaction patterns, including planning, coordination, and task handoff
• Implement agent orchestration using frameworks such as LangChain, LangGraph, CrewAI, or similar
• Integrate self-hosted LLMs as well as commercial LLM APIs
• Develop backend services in Python, including:
• Agent orchestration and execution logic
• Tool integrations and external API access
• Backend APIs for frontend consumption
• Develop and maintain frontend components using React.js, including:
• User interfaces for interacting with AI agents
• Visualisation of agent outputs and intermediate results
• Human-in-the-loop workflows (review, approval, override)
• Implement agent evaluation and monitoring, covering:
• Response quality and correctness
• Performance and latency
• Cost and resource usage
• Work closely with business users across departments to understand workflows and translate business needs into agent-based solutions
• Produce technical documentation, including system design, agent patterns, and usage guidelines
Job Requirements
• Diploma or higher qualification in Artificial Intelligence, Data Science, Computer Science, or a related field
• Strong hands-on experience in multi-agent AI system design and implementation
• Proficient in Python (backend) and React.js (frontend)
• Practical experience with LangChain, LangGraph, CrewAI, or similar AI agent frameworks
• Experience working with both self-hosted LLMs and commercial LLM services
• Familiarity with agent evaluation techniques and AI quality assessment
• Solid understanding of LLM limitations, including hallucinations and non-deterministic behavior
• Ability to design reliable AI systems with appropriate guardrails and human-in-the-loop controls
• Strong communication skills and ability to engage business users with varying technical backgrounds
• Able to work independently while collaborating closely with internal teams
