AI Agent Security Research Engineer (Web3)

TL;DR

AI Agent Security Research Engineer (Web3): Designing and implementing multi-agent collaborative systems for code auditing and AI system protection with an accent on vulnerability detection, prompt injection defense, and LLM security. Focus on building scalable Agent service architectures, integrating DevSecOps workflows, and ensuring the observability and traceability of agent actions in production environments.

Location: Must be based in Hong Kong or Singapore

Company

hirify.global is a leading global crypto exchange and developer of the hirify.global Wallet, providing millions of users with access to decentralized applications and trading markets.

What you will do

• Design and implement multi-agent collaborative systems for code auditing, including vulnerability detection and sensitive data leakage prevention.
• Integrate RAG, Chain-of-Thought, and reflection technologies into security agents to optimize detection accuracy.
• Develop security protection frameworks for LLM applications, covering input, output, and runtime layers.
• Build automated workflows for threat response, alert classification, and SOAR platform integration.
• Construct highly available, scalable Agent service architectures for large-scale concurrent scanning tasks.
• Ensure observability and traceability of agent actions in production, adhering to industry standards like OWASP Top 10 for LLMs.

Requirements

• 3+ years of backend development experience using Python, Go, or Java.
• Hands-on experience deploying LLM Agents in production with knowledge of hallucination handling and tool invocation fault tolerance.
• Proven experience with AI security, including defense strategies against prompt injection, jailbreaking, and malicious agent injection.
• Proficiency in at least one agent framework such as LangChain, LlamaIndex, AutoGen, or LangGraph.
• Expertise in Docker, Kubernetes, and microservices architecture design.
• Must be based in or able to work from Hong Kong or Singapore.

Nice to have

• Experience with SAST/SCA tools like CodeQL, Semgrep, or SonarQube.
• Experience with LLM fine-tuning (SFT, LoRA) or local model optimization.
• Track record of open-source contributions or security competition awards (CTF, CVE/CNVD).

Culture & Benefits

• Competitive total compensation package.
• Comprehensive healthcare schemes for employees and dependants.
• L&D programs and education subsidies for professional growth.
• Wellness and meal allowances.
• Collaborative team environment focused on core principles of integrity and execution.