LLM Security Engineer

Arcitix Technologies; San Francisco (CA); On-site; USD 120,000 - 160,000; Full time

Job summary

Arcitix Technologies in San Francisco is looking for an LLM Security Engineer to design and implement security measures for AI applications.

The position involves building defenses against prompt injection, output moderation, and authorization for agentic systems. You will need 4+ years of experience in software or security engineering and strong skills in Python.

This role is critical to the security aspects of customer AI applications, focusing on effectiveness and efficiency.

Qualifications

• 4+ years in software or security engineering.
• Experience with low-latency, high-throughput services is a plus.

Responsibilities

• Build inline guardrails for prompt-injection and jailbreak defense.
• Implement output moderation and PII/secret leakage prevention.
• Design tool-call authorization for agentic and RAG systems.

Skills

Strong Python and API-design skills

Hands-on experience building LLM or agent applications

Solid grasp of LLM threat models and guardrail patterns

Education

Bachelor's degree in CS or equivalent practical experience

Job description

Design and ship the guardrails behind our LLM firewall — prompt-injection filtering, output moderation, and agent authorization.

About the role

LLM Security Engineers own the guardrails behind our LLM firewall. You will build inline defenses against prompt injection and jailbreaks, design output moderation and leak prevention, and implement tool-call authorization for agentic systems.

This role sits at the intersection of product and security engineering — you ship code that runs in the critical path of customer AI applications, so correctness, latency, and false-positive rates all matter.

Key Responsibilities

• Build inline guardrails for prompt-injection and jailbreak defense
• Implement output moderation and PII/secret leakage prevention
• Design tool-call authorization for agentic and RAG systems
• Integrate guardrails with client RAG and agent pipelines
• Benchmark guardrail efficacy against live red-team findings

Requirements

• Bachelor's degree in CS or equivalent practical experience
• 4+ years in software or security engineering
• Hands-on experience building LLM or agent applications
• Strong Python and API-design skills
• Solid grasp of LLM threat models and guardrail patterns

Nice to have

• Experience with low-latency, high-throughput services
• Familiarity with content-safety classifiers and embeddings
• Prior work on developer-facing security products