The opportunity
Crusoe is on a mission to accelerate the abundance of energy and intelligence . As the only vertically integrated AI infrastructure company built from the ground up, we own and operate each layer of the stack — from electrons to tokens — to power the world's most ambitious AI workloads.
What you'll do
Detection & Intelligence Ownership: Own the full detection stack — heuristics, threshold calibration, precision/recall tuning, and the rule systems that define what "something is wrong" means for the platform.
Anomaly Detection Pipelines: Design and maintain detection systems including straggler node detection, GPU health signals, and fleet-level behavioral baselines.
Signal Calibration: Drive detection fidelity by reducing false positives, increasing signal coverage, and building feedback loops that keep thresholds accurate as the fleet grows.
ML/RL Integration: Evaluate and integrate machine learning and reinforcement learning techniques where they outperform rule-based approaches — and know when not to reach for a model.
Product Engineering: Ship customer-facing features end-to-end across the CCIA stack — alert rule engine, control plane APIs, automated action systems, and insights delivery surfaces.
to-1 & Scale: Build new systems from scratch and scale existing ones to support Crusoe's rapidly growing global fleet.
What they're looking for
- Cross-Functional Collaboration: Work closely with product counterparts to shape requirements early and partner with the data science team to develop and validate detection models.
- System Design: Participate in design discussions across teams, contribute architectural perspective, and help evaluate technical trade-offs.
- Technical Mentorship: Mentor engineers at all levels through code review, design feedback, and direct coaching, and contribute to hiring by helping define what great looks like.
- Anomaly Detection & Heuristics Expertise: Deep experience building anomaly detection systems, heuristics-based rule engines, or ML/RL systems for infrastructure or data-intensive domains.
