Description

Tel Aviv

• Hybrid | Full-Time | Cybersecurity
  
  

What You'll Do

• Design and develop LLM-powered security features and internal AI tools — RAG pipelines, multi-agent workflows, prompt-engineered systems for cybersecurity
• Architect and operate multi-agent systems in production — orchestration, inter-agent communication, task delegation, failure handling at scale
• Build agent monitoring and observability pipelines — tracing, drift/failure detection, alerting, reliability SLAs
• Build and maintain scalable MLOps infrastructure — model serving, eval frameworks, experiment tracking, CI/CD for ML
• Fine-tune and adapt foundation models on internal datasets (network telemetry, security logs, threat intel)
• Establish best practices for model observability, safety, and responsible AI deployment
• Stay current with the LLM/GenAI ecosystem; drive updates to the AI SDLC and AI Research cycle
  
  

tions 'from scratch'

Requirements

Must-Have:

• 5–8 years SWE (2–3 in AI/ML)
• Production LLM apps (RAG/agents/tool-use/fine-tuning)
• Production multi-agent systems
• Agent observability
• LangChain/LangGraph/Bedrock AgentCore
• Strong Python
• MLOps pipelines
• Transformers/embeddings/vector DBs
• Cloud + K8s.
  
  

Nice-to-Have

• Cybersecurity background (significant plus)
• Networking (SDN/BGP)
• Model eval (LLM-as-judge/RAGAS)
• MCP
• Telecom/enterprise SaaS
• publications/OSS in GenAI.
  
  

Stack

Python, PyTorch, OpenAI/Anthropic APIs, LangChain, LangGraph, AWS Bedrock AgentCore, LangSmith, Kubernetes, Kafka, Elasticsearch, AWS, PostgreSQL