Software Engineer, AI Safety & Systems

Realm Labs•Sunnyvale, CA

29d

About The Position

We are hiring a Founding AI Safety & Systems Engineer to build the core technical systems that detect, analyze, and mitigate AI risks in production. This role sits at the intersection of Software engineering, ML systems, and AI safety. You will design and implement systems that inspect model behavior, process large-scale signals, and enforce safety and governance policies in real time. This is not a research-only role. You will focus on turning safety ideas into robust, scalable production systems that enterprises can trust. This role is ideal for an engineer who understands how modern AI systems work, cares deeply about correctness and robustness, and wants to shape how AI safety is implemented in the real world. About Realm Labs Realm Labs is an AI trust and security startup. We help enterprises detect, debug, and prevent AI misbehavior in production through deep inspection, observability, and runtime enforcement. Our systems operate in high-stakes environments where correctness, explainability, and reliability matter.

Requirements

5+ years of professional software engineering experience
Strong backend or systems engineering background
Experience building production systems with strict correctness requirements
Ability to reason about complex system behavior and edge cases
Comfortable working with ML-adjacent systems, even if not a researcher

Nice To Haves

Experience working on AI safety, trust, or reliability systems
Familiarity with LLMs, model inference pipelines, or ML evaluation systems
Experience processing large-scale structured and unstructured data
Strong interest in AI governance, safety, or responsible deployment
Startup experience or ownership of greenfield systems

Responsibilities

Build systems that analyze and evaluate AI behavior at runtime, including: Input and output inspection Intermediate signal processing Policy and risk evaluation pipelines
Design and implement backend services that support: Real-time moderation and safety decisions High-throughput model post-processing Secure, auditable decision paths
Work closely with ML and product teams to: Translate safety requirements into system designs Integrate models, probes, and classifiers into production Debug model and system behavior end-to-end
Implement scalable data pipelines using: Python, Go, Rust, and C++ PostgreSQL, document databases, Elasticsearch, and similar systems
Build infrastructure that emphasizes: Correctness over heuristics Clear failure modes and graceful degradation Strong observability and explainability
Establish best practices for: Evaluating and validating safety systems Versioning and rolling out safety logic Monitoring regressions and edge cases