高级AI开发工程师（安全）-剪映CapCut（广州/深圳）
Core
Design and develop the overall architecture for CapCut's AI safety gateway service, ensuring low latency, high throughput, and high availability to support real-time calls for hundreds of millions of users.
Role type
Senior IC backend engineer (AI safety/LLM guardrails)
Builds
LLM safety guardrails, prompt/output filtering, jailbreak detection, and sensitive information identification/desensitization pipelines.
Domain
Generative AI security, content safety, distributed systems
Deliverable
production ML models
Required skills
Go, Java, C++, distributed system design, microservices architecture, gRPC/HTTP, caching, message queues, database management, LLM principles, adversarial attack perspective
Preferred skills
Python, cloud-native (Kubernetes, Istio, Linkerd), observability, large-scale performance optimization, TensorFlow, PyTorch
Responsibilities
Design flexible, pluggable safety pipelines integrating multiple detection models and rule engines; collaborate with research teams to design defense mechanisms against emerging attack vectors; build automated evaluation frameworks to monitor and quantify safety performance; develop backend tools for security policy configuration and risk analysis.
Seniority
Senior, hands-on IC
