Applied AI Engineer, Autonomous Defense
Horizon3 · Remote · US, Remote ·
COMPENSATION
€•• – ••k
Listed on the employer's job board (Ashby)
- Fully remote (US)
- AI + Cybersecurity
- Python + PyTorch
- Senior level
- $313k-$369k base
Join Horizon3 to develop AI agents that transform offensive security knowledge into automated defensive actions. You will build reasoning systems that map attack paths to specific control changes, such as firewall rules or IAM policies, and verify their effectiveness by re-running attacks.
The team works on closed-loop defense at enterprise scale, using real outcome signals from hundreds of thousands of prior tests. You will design safety architectures for autonomous changes and collaborate with attack engineers to ensure reliability in production environments.
What you'll do
- Build reasoning systems mapping attack paths to ranked control changes
- Convert pentest data into training and evaluation datasets
- Design counterfactual experiments in representative test environments
- Create reasoning layers for heterogeneous vendor APIs and policy models
- Implement safety architecture including dry-runs and rollback mechanisms
- Collaborate with engineers to diagnose agent failure modes
- Own end-to-end problem solving in ambiguous, fast-moving contexts
What you'll need
- Master's in CS/ML or equivalent plus 4+ years engineering experience
- Strong ML engineering experience with deep learning and PyTorch
- Experience with post-training LLMs or robust agentic systems
- Track record building evaluation systems for open-ended tasks
- Proficiency in Python for production-quality code
- Experience with data pipelines, distributed systems, and AWS
- Ability to reason over messy real-world data like logs and configs
- Strong communication skills and ability to research unfamiliar systems
Nice to have
- Background in detection engineering, purple teaming, or IR
- Familiarity with security control planes like CrowdStrike or Palo Alto
- Experience with causal inference, graph reasoning, or Neo4j
- Knowledge of building automation with safety and change management
- Experience with adversarial robustness or prompt injection
- Background integrating ML into multi-tenant SaaS or air-gapped envs
Stack
- Python
- PyTorch
- AWS
- Neo4j
- CrowdStrike
- SentinelOne
- Palo Alto
- Entra ID
- Splunk
How you'll work
- Remote and hybrid models available depending on location
- Some roles may require presence in Chicago office
- Fast-paced environment with ambiguous requirements
What you get
- Base salary range $313,000 - $369,000 annually
- Equity package via stock options
- Health, vision, and dental insurance for family
- Flexible vacation policy
- Generous parental leave
- Inclusive culture with growth opportunities