YC W24 startup building AI-powered video management systems for law enforcement Real-Time Crime Centers and transportation departments. We help agencies search camera networks in natural language and detect and respond to threats in real time. Tech Stack Frontend: Next.js, React, TypeScript, gRPC-Web Backend: Nest.js, REST, Protocol Buffers, gRPC AI/ML: Python, PyTorch, TensorRT GPU: Custom CUDA kernels for video decoding and multi-stream inference Infrastructure: RabbitMQ, Bazel, Docker, Systemd Data: Vector database, PostgreSQL-17 Video: ONVIF, RTSP, HLS, WebRTC What You'll Build Vantage VMS features: natural language video search, multi-camera playback, incident reporting Vehicle detection and tracking pipelines for DOT traffic analysis Real-time alert systems for suspect/vehicle identification ONVIF camera discovery and integration services Analytics dashboards showing detection accuracy, system health, and search performance Example Projects Multi-camera grid view with synchronized timeline playback CJIS-compliant audit logging: track all user actions, video exports, and system access Camera auto-discovery via ONVIF with health monitoring and alerting Video export with chain of custody: watermarking, MD5 hashing, evidence lock Storage management system with retention policies and auto-overwrite Role-based access control (Admin/Operator/Viewer) with session management Recording scheduler: continuous, motion-based, or time-based per camera Bookmarking and evidence tagging for investigation workflows