LLM Safety Engineer: Backend & Guardrails

LINKTRIX Consultants, Asia Pacific; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

LINKTRIX Consultants, Asia Pacific is seeking an AI Engineer to conduct safety testing of Large Language Models, develop safety guardrails, and support the deployment of safe AI applications. You will build Python-based backend services and APIs, optimise performance, and collaborate with product teams on safety integration.

The role emphasizes hands-on testing, guardrail development (NeMo / LLM Guard), and proactive learning in AI safety and responsible AI practices.

Qualifications

• Experience in Python backend development and API design.
• Experience with cloud deployment, preferably AWS.
• Strong understanding of LLMs, prompt engineering, and LLM-based app development.
• Interest in AI safety and responsible AI with proactive learning.

Responsibilities

• Conduct safety testing of LLMs and identify vulnerabilities.
• Develop safety guardrails using open-source frameworks.
• Build, deploy, and optimise LLM-based applications with model fine-tuning for safety alignment.
• Develop backend services and APIs using Python with performance focus.
• Collaborate with AI product teams to integrate safety testing and guardrails.
• Promote responsible AI through experimentation and knowledge sharing.

Skills

Python backend

API development

ML performance

Cloud deployment

AWS

LLMs understanding

Prompt engineering

Safety testing

Guardrails

Performance optimisation

Collaboration

Tools

NeMo

LLM Guard

Job description

LINKTRIX Consultants, Asia Pacific is seeking an AI Engineer to conduct safety testing of Large Language Models, develop safety guardrails, and support the deployment of safe AI applications. You will build Python-based backend services and APIs, optimise performance, and collaborate with product teams on safety integration.

The role emphasizes hands-on testing, guardrail development (NeMo / LLM Guard), and proactive learning in AI safety and responsible AI practices.