Principal Engineer - AI

Safe Security•San Jose, CA

16h•$250,000 - $300,000

About The Position

At SAFE Security, our mission is bold and ambitious: We Will Build CyberAGI â a super-specialized system of intelligence that autonomously predicts, detects, and remediates threats. This isnât just a visionâitâs the future weâre building every day, with the best minds in AI, cybersecurity, and risk. At SAFE, we empower individuals and teams with the freedom and responsibility to align their goals, ensuring we all move towards this goal together. We operate with radical transparency, autonomy, and accountabilityâthereâs no room for brilliant jerks. We embrace a culture-first approach, offering an unlimited vacation policy, a high-trust work environment, and a commitment to continuous learning. For us, Culture is Our Strategyâcheck out our Culture Memo to dive deeper into what makes SAFE unique. As a Principal Engineer - AI, you will define and lead the technical direction of AI systems that power Safeâs CRQ, CTEM, and TPRM products, including agentic workflows, RAG pipelines, LLM orchestration, and AI-native developer tooling. Youâll be the hands-on architect behind Safeâs AI engineering stack, bridging model intelligence with production-grade infrastructure. Youâll collaborate with product, data, and platform teams to design scalable, explainable, and enterprise-ready systems. This is a high-impact, technical leadership role that will shape how AI is built, deployed, and governed across Safe.

Requirements

Experience: 12+ years total experience in software engineering, including 4+ years building AI/ML systems or large-scale data/LLM infrastructure.
Core Technical Skills:
- Strong programming fundamentals in Python, Go, or TypeScript
- Deep understanding of LLM-based architectures, prompt engineering, and RAG pipelines
- Hands-on experience with LangChain, LlamaIndex, or equivalent orchestration frameworks
- Vector databases (FAISS, Pinecone, Weaviate, Redis Vector, or Milvus)
- Cloud model deployment (AWS SageMaker, Bedrock, Vertex AI, or custom inference APIs)
- Data systems: Snowflake, Iceberg, S3, Postgres/MySQL
MLOps & Infra: Familiar with model versioning, CI/CD for ML, and performance optimization for real-time inference.
Applied AI Focus: Practical understanding of evaluation metrics, hallucination detection, RAG reliability, and enterprise AI safety.

Nice To Haves

Experience integrating AI into cybersecurity or risk management products
Familiarity with multi-agent systems and autonomous workflows (CrewAI, LangGraph, AutoGen)
Experience building AI evaluation dashboards and AI observability stacks
Knowledge of knowledge graphs, semantic search, or retrieval pipelines
Exposure to data governance, compliance, or SOC2/ISO 27001 environments
Published research, open-source contributions, or prior leadership of AI teams is a strong plus

Responsibilities

Architect Safeâs AI Systems: Design and scale AI-driven components â LLM orchestration, retrieval-augmented generation (RAG), vector stores, prompt pipelines, and AI microservices. Drive architecture for AI observability, safety, and evaluation (precision, recall, F1, hallucination detection, cost metrics).
Productionize AI Agents: Build multi-turn, goal-oriented agent systems that automate reasoning across TPRM, CTEM, and CRQ domains (e.g., control reviews, issue RCA, automated responses). Ensure reliability, traceability, and deterministic behavior in production.
AI Infrastructure & Platform Ownership: Partner with Platform & DevOps teams to operationalize model serving (AWS SageMaker, Bedrock, or self-hosted Llama), build AI APIs, and manage model lifecycle and versioning. Establish feature stores, embedding management, and in-memory retrieval layers.
Data Pipeline & Knowledge Graph Integration: Work with Data Engineering to design pipelines for structured and unstructured data ingestion, semantic indexing, and context retrieval (Snowflake + Iceberg + LlamaIndex).
AI Evaluation, Monitoring & Governance: Define internal frameworks for golden dataset validation, LLM evaluation (LangFuse/LangSmith), and safety enforcement policies. Implement human-in-the-loop (HITL) mechanisms and continuous feedback loops.
Mentor & Multiply: Guide AI and backend engineers on architectural design, experimentation methodologies, and prompt optimization. Collaborate with product leaders to translate abstract AI goals into measurable engineering deliverables.