Agentic AI Engineer (MY)

Oxydata Software Sdn Bhd

Location: Petaling Jaya, Selangor Work Mode: Hybrid Contract Type: Full-time Employment Eligibility: Malaysian Only

About Oxydata

Oxydata Software Sdn Bhd is a Malaysia Digital-certified AI and data engineering company founded in 2006. We design and deploy production AI systems for enterprise clients across financial services, human resource, energy, and telecommunications. We are now expanding our Agentic AI practice — building autonomous, multi-step AI workflows that go beyond single-prompt responses to deliver real business process automation.

The Role

We are looking for an Agentic AI Engineer to lead the design and implementation of multi-agent AI systems across our product and client delivery portfolio. This role is not about building dashboards — it is about architecting autonomous AI workflows that reason, plan, use tools, and execute multi-step tasks with minimal human intervention. You will work in a small, senior team using AI-assisted development tools as standard practice.

What You'll Build

Multi-Agent Orchestration

• Design and implement multi-agent pipelines using LangGraph and CrewAI
• Agent role definition, task decomposition, inter-agent communication, and state management
• Supervisor/worker agent architectures for complex, multi-step reasoning tasks

Workflow Automation

• Build production automation workflows using N8N — triggers, conditionals, API calls, data transforms
• Integrate AI agents into broader business process automation pipelines
• Design human-in-the-loop checkpoints for workflows requiring approval or escalation

Tool Use & Function Calling

• Implement tool-calling agents with access to APIs, databases, search, and file systems
• Build custom tools and MCP (Model Context Protocol) servers for agent consumption
• Memory management — short-term, long-term, and episodic memory for persistent agents

Multimodal AI

• GPT Vision integration for document parsing, screenshot analysis, and visual data extraction
• Multimodal pipelines combining text, image, and structured data inputs

RAG & Knowledge Retrieval

• Integrate RAG pipelines as retrieval tools within agentic workflows
• Vector search (pgvector, Pinecone, or similar) for agent knowledge grounding
• Prompt engineering for agent personas, reasoning chains, and output formatting

Production Deployment

• Deploy agentic systems on self-hosted VPS infrastructure
• Observability, logging, and failure handling for long-running agent workflows
• API endpoints (FastAPI) exposing agent capabilities to frontend and external systems

Must-Haves

• Education: Bachelor's in Computer Science, Software Engineering, AI, or related field
• Experience: 1–4 years in AI/ML engineering with hands-on production experience in agentic or LLM-based systems
• Agentic Frameworks: Demonstrated production experience with LangGraph and/or CrewAI
• Workflow Automation: Hands-on N8N workflow design and deployment
• Multimodal AI: GPT Vision or equivalent for image/document understanding
• Language: Python (primary); comfortable reading JavaScript
• LLM APIs: OpenAI API, Claude API, function calling, tool use, structured outputs
• Prompt Engineering: System prompts, chain-of-thought, ReAct patterns, output formatting
• RAG Fundamentals: Embeddings, vector search, retrieval grounding — as agent tools
• Backend: FastAPI for exposing agent capabilities as APIs
• DevOps: Git, VPS/cloud deployment, basic Linux administration, error handling for async workflows

Nice-to-Haves

• LangChain or LlamaIndex for knowledge-base construction
• MCP (Model Context Protocol) server development
• Experience with AI-assisted development tools (Cursor, GitHub Copilot)
• Docker for containerised agent deployment
• CI/CD pipelines for agent workflow testing and deployment
• React.js — basic frontend ability to build simple agent UIs or dashboards
• Experience with structured output validation (Pydantic, instructor library)

What We Offer

• Be among the first engineers in Malaysia building production-grade Agentic AI systems
• Work on real client deployments — not internal tools or proofs of concept
• AI-first development culture — Cursor and Claude are standard tools, not novelties
• Small, senior team — high ownership, direct impact, no bureaucracy
• Competitive compensation commensurate with experience

How to Apply

Submit your CV via our careers page: https://oxydata.ai/careers