Senior AI Engineer

 

About Radiant Security

Radiant Security is building the most advanced AI SOC platform, featuring unbounded alert triage, investigation, and response for security teams at scale. Our platform ingests alerts from across an organization's entire security stack (SIEM, EDR, identity, cloud) and uses AI to triage, investigate, and surface what actually matters. We're replacing alert fatigue with clear signal, so analysts can focus on real threats.

We're a small, fast-moving team. We ship continuously, stay close to customers, and hold ourselves to a high standard. Our product touches the daily workflows of security teams, and decisions we make have a direct impact on how quickly threats get resolved.

Join us and boost your career with hands-on AI experience.

The Role

Cybersecurity generates more signal than any human team can process. The AI Engineer's job is to change that.

In this role, you'll build the agentic systems that turn overwhelming alert volume into clear, confident decisions — pipelines where AI agents collaborate, reason, and escalate the way the best analysts do, but at machine speed and scale. This isn't prompt experimentation in a notebook. You'll own the full lifecycle of production pipelines: the retrieval systems that surface the right context, the routing logic that sends data where it needs to go, the DAG that encodes the decision structure of a real security problem, and the agents themselves that reason over it all.

Every layer is yours. Every improvement is measurable. And the problems you're solving — faster triage, fewer missed threats, less analyst burnout — are ones that matter.

What you’ll do:

DAG Design & Pipeline Orchestration

• Design and implement the DAG-based pipeline architecture. Decompose complex cybersecurity workflows into well-scoped pipeline stages and define data contracts between them.

Context Retrieval & Data Routing

• Build the retrieval and enrichment systems that feed agents the context they need: structured lookups, vector search, tool calls, API integrations, or database queries. Design routing logic that directs incoming data to the appropriate pipeline entry point or agent branch. Ensure that every agent in the pipeline receives exactly the context it needs.

Agent Development & Prompt Engineering

• Design and build individual agents with well-scoped, production-ready prompts tailored to specific cybersecurity tasks (e.g., alert classification, entity extraction, severity scoring, false positive filtering). Apply systematic prompt optimization techniques, iterating based on measurable outcomes rather than intuition.

LLM Lifecycle Management

• Evaluate new or alternative LLMs using structured methodologies and project-specific metrics when a model substitution is warranted (e.g., upgraded model version, cost optimization, latency requirements).

Production Operations & Evaluation

• Monitor pipeline performance end-to-end in production — latency, accuracy, cost, failure modes, and per-stage quality — and drive continuous improvement across all layers. Build and curate labeled datasets from production telemetry for evaluation, regression testing, and fine-tuning purposes.

Things we’re looking for

Required Qualifications

• Strong programming skills in Python; 5 years of experience building and deploying production services end-to-end.
• Hands-on experience with LLM APIs (e.g., OpenAI, Anthropic, Gemini) and prompt engineering techniques.
• Experience designing and implementing DAG-based workflows or pipeline orchestration systems.
• Familiarity with data retrieval patterns: vector search, structured queries, API integrations, and context assembly for LLM consumption.
• Ability to work with unstructured or semi-structured data in a security or operational context.
• Clear written communication — pipeline design decisions, agent behavior, and experimental results should be well-documented.

Preferred Qualifications

• Background in cybersecurity, SOC operations, or threat intelligence.
• Experience with evaluation frameworks (e.g., RAGAS, LangSmith, custom harnesses).
• Familiarity with fine-tuning, RLHF, or model distillation workflows.
• Exposure to agentic frameworks (e.g., LangGraph, CrewAI, AutoGen, or custom implementations).
• Solid understanding of statistical testing and experimental design (A/B testing, significance testing, confidence intervals).

The process

We’re a startup and we make decisions quickly. Our process is designed to give you the best glimpse of our team and allow us to evaluate your technical and culture fit.

Application Review > Recruiter Screening > Hiring Manager Interview > Technical Interviews > Virtual Onsite