(30378) Applied AI Engineer, NUSX Labs

• An integration backbone. Every system plugs in the same way: one auth pattern, one schema contract, contract tests, and docs a new teammate can copy.
• An agent harness with deterministic guardrails. A tool server that lets agents and automations write to shared boards, with the limits enforced in code rather than the prompt: allow-listed boards and columns, every write logged, no deletes.
• An intake agent. An inbound email arrives, and a structured, classified ticket comes out the other side, backed by an evaluation set so we know exactly how often it gets it right.
• A milestone watchdog. It reads award milestones from the source systems, spots what is due or at risk before anyone has to chase, and writes a status flag back for the team.
• A demo you can stand behind in front of senior leadership, with the documentation to hand it on.
  

• You own it end to end. From day one, you build, test, ship and run each connector or workflow yourself.
• Your work ships. It goes to real users, not an internal tool, not a shiny demo.
• Close to users. You run requirements and acceptance testing directly with the people who use the result, and you change the work based on what you learn.
• Close to the decisions. You work directly with the technical lead and present your work to senior leadership.
• AI-native by default. Claude Code or an equivalent is part of every working day, and we expect you to be fast and surgical with it.

• At least 3 years of hands-on experience in software engineering, data engineering, system integration or automation development. An AI-specific job title is not required.
• Strong Python or TypeScript/Node, with experience building and consuming REST and/or GraphQL APIs, authentication (OAuth, service accounts) and SQL/Postgres.
• You have shipped at least one feature, integration or workflow that calls an LLM API. Classification, extraction, agents and RAG all count, and no particular framework is expected.
• Good testing habits, including integration or contract testing, and documentation another person can follow.
• Daily use of AI-assisted coding tools (Claude Code, Cursor, GitHub Copilot or similar).
• Comfort working with non-technical users and turning a vague pain point into working software.
• Degree or diploma in a technical discipline, or equivalent practical experience.
  

• Familiar with writing MCP servers from scratch or other tool-calling backends.
• Familiar with one or more of: Monday.com (GraphQL), n8n, Google Workspace or Slack integrations.
• Familiar with agentic skill writing, optimisation.
• Evaluation sets or prompt regression checks for LLM features.
• Embeddings, vector databases (pgvector or similar) and RAG.
• Experience with Google Cloud.
• Comfortable reading and working in English and Chinese.

Job Type: 2-Year Contract

Location: Kent Ridge Campus

Organisation: NUSX

Department: Administration

Job requisition ID: 30378