platoseedplatoseed
Low-false-positive intrusion detection for remote outdoor sites
Rimward helps operators of solar farms and other remote outdoor infrastructure detect intrusions from existing camera networks. We turn existing CCTV into operator-ready alerts with video verification, so security teams can reduce false alarms, respond faster, and avoid heavy hardware retrofits. We are focused on sites where recurring theft, perimeter intrusion, and low staffing make traditional monitoring noisy and operationally weak. Our approach is software-heavy, field-oriented, and built to work with real site constraints.
Rimward provides intrusion alerts layered on top of existing CCTV for supervision teams at remote outdoor sites, focusing on reducing false positives with a measured budget and audit-ready evidence. It works without rip-and-replace and emphasizes faster, more reliable decisions on sites like solar farms.
Rimward connects read-only to current CCTV via RTSP or existing VMS and adds a software layer for intrusion detection and video verification. Each alert includes a clip, zone, timestamp, and classification reason, then goes through a 5-step process: Detect, Verify, Decide, Label, and Report. Alerts are tuned against a site-level false-positive target, with nightly site reliability reports and a weekly tuning review. The system operates on top of the existing monitoring stack and starts with pilots using a small camera set, agreed schedules, and an evidence package for rapid decision-making.
Who it’s for: Solar farms and other remote outdoor technical sites with existing camera coverage and internal supervision
Live pilot program in France; emphasis on remote outdoor sites; references to partnerships with solar/IPP contexts and ongoing tuning and reliability reporting
FG is the founder & CEO of Rimward (S23). He earned a MSc in Mathematics and Computer Science from CentraleSupélec and has a deep background in ML, having led the research team at Mindee (YC W21), published several papers, and become one of the top PyTorch contributors. Beyond AI, FG specializes in large-scale engineering systems - he built and deployed a ground-based wildfire detection network across Europe, and South America starting with a Raspberry Pi, a camera and his laptop.
Bridging open-source efficiency to drive enterprise excellence.
Quack AI launches a dual-interface solution for large software companies to streamline OSS-style contribution processes: a web maintainer interface to design, curate, and sequence contribution steps, and a VSCode extension to help developers identify the right issues and align with contribution flows, aiming to enable cross-team participation in pivotal projects.
From the original launch (Aug 2023) — may be outdated.
Formerly “Quack AI”, “Sourcepulse” · why startups rename →
Streamlining Data Capture to Report Creation for Heavy Industries
Detect cyber attacks on critical physical infrastructure
