Fleet Reliability Engineer

Job description

**Company Background**Specter's mission is to help automate the physical world. Today, we build video sensors with state-of-the-art AI agents that answer any question, anywhere in their environments. Our systems can automatically detect and reason about any physical activity captured on camera, from security incidents (e.g. perimeter intrusion, theft, LPR), to safety monitoring (e.g. PPE detection, injured people), to operational efficiency (e.g. material tracking, congestion monitoring). We offer both long range wireless (1km range) and wired sensor variants to suit any deployment. Our co-founders Xerxes and Philip are passionate about empowering our partners in the fast approaching world of physical AI and robotics. We are a small, fast growing team who hail from Anduril, Tesla, Uber, and the U.S. Special Forces. **The Role**We’re hiring a Fleet Reliability Engineer to keep our sensor fleet running in the field by building the data, analytics, and recovery mechanisms that prevent failures from becoming incidents. As we scale toward thousands of sensors, fleet health becomes a data-and-systems problem. This is the proactive, highest-leverage side of reliability: own the telemetry and data pipeline, verify that fixes hold fleet-wide, and turn field signal into cost-weighted decisions about what to fix first. Much of today’s operational load is addressable through better instrumentation, alert hygiene, and recovery verification, at little to no field cost. ## **Responsibilities:** Reliability Data Platform — Primary - Own the fleet’s reliability data pipeline end to end: telemetry aggregation, storage, and instrumentation. - Drive down observability cost — own the tooling spend and cut what we pay for but don’t use. - Instrument the fleet and own the health metrics that measure reliability. Proof-of-Recovery & Alert Hygiene - Verify that fixes hold fleet-wide, not just on the device that paged. - Cut alert noise at the source — separate real failures from self-resolving ones. - Turn repeat failure patterns into automated detection and recovery. Fleet Health & Failure-Mode Analytics - Turn fleet telemetry into a live picture of which cohorts, hardware revisions, and firmware versions are trending toward failure, and why. - Build the failure-mode analysis that tells engineering what to fix at the source. - Own fleet-wide trend and forecasting work, including power and solar planning. Reliability Economics & Prioritization - Score reliability work in dollars — field-trip cost, hardware-return cost, observability spend — and prioritize the most expensive problems first. - Set and track the fleet’s reliability targets: uptime, offline rate, truck-rolls per sensor-year. - Give the team the data to make reliability-versus-cost tradeoffs. ## **Qualifications:** - Strong data and software skills — Python (or Go) and SQL — and the ability to own a data pipeline end to end. - Hands-on building and tuning observability stacks (OpenTelemetry, Grafana, Prometheus, Datadog, or similar), including their cost. - Experience operating physical or embedded device fleets at scale, and reasoning about how hardware fails in the field. - Comfortable turning messy field telemetry into trends, failure modes, and forecasts. - Fluency with databases and data modeling (PostgreSQL or equivalent); infrastructure-as-code familiarity (Terraform or similar) a plus. - Bias toward building mechanisms over doing manual work. - Nice to have: reliability/SRE fundamentals (SLOs, error budgets, proof-of-recovery) applied to a physical fleet. - Nice to have: experience across the hardware-software boundary — power, connectivity, and physical failure modes.