Senior Agentic AI Evaluation Engineer

RESARO SINGAPORE PTE. LTD.; Singapore; On-site; SGD 120,000 - 170,000; Full time

Job summary

Resaro is a Singapore-based independent AI assurance company building the Approved Intelligence Platform (AIP) to rigorously evaluate AI systems for government, defence, and regulated sectors. We seek a senior engineer to own the agentic-evaluation engine and guide the testing framework used to verify tool-using, multi-step agents and their safety across missions.

You will work hands-on, own the technical vertical, and collaborate with Product to shape the roadmap while extending evaluations to

Qualifications

• 5+ years of professional software engineering experience.
• Experience with agentic AI evaluation and tool-calling.
• Strong discipline in measuring and benchmarking AI systems.

Responsibilities

• Own the agentic-evaluation engine and build the scoring capability for important metrics.
• Evaluate stateful, multi-turn behaviour and memory across sessions.
• Design for evaluator integrity, isolate runner from agent, and audit ground truth.
• Develop agentic-security surface including red-teaming and permission-inheritance checks.
• Close the lab-to-production gap by surface areas where demos fail in real deployments.
• Turn governance requirements (EU AI Act, NIST AI RMF, ISO 42001) into testable evaluation logic.
• Set the technical bar via spikes, ADRs, and code reviews for reproducibility.
• Extend evaluation to embodied/autonomous systems as needed.

Skills

Python

Agentic AI

Production systems

Software engineering

Tools

Kubernetes

Helm

CUDA

Job description

Resaro is a Singapore-based independent AI assurance company building the Approved Intelligence Platform (AIP) to rigorously evaluate AI systems for government, defence, and regulated sectors. We seek a senior engineer to own the agentic-evaluation engine and guide the testing framework used to verify tool-using, multi-step agents and their safety across missions.

You will work hands-on, own the technical vertical, and collaborate with Product to shape the roadmap while extending evaluations to