The edge AI infrastructure you don't have to build from scratch
NEPI handles roughly 90% of what most smart systems need out of the box. Teams customize the remaining 10% for their specific application.
The problem
Most teams spend months on infrastructure they never wanted to build
If your team has ever started a sensor integration project and spent three months writing plumbing before touching the actual problem - there is a better starting point.
Building a smart system today means stitching together hardware drivers, AI pipelines, automation logic, data collection, and a monitoring interface. That is months of work before you get to the application your team was hired to build.
Instead of building a custom sensor integration layer, an AI pipeline, and a data collection system from scratch, teams build on NEPI.
Hardware drivers
Plug-and-play support for cameras, LiDAR, sonar, pan/tilt, and actuators. NEPI's standardized abstraction layers (IDX, PTX, LSX, RBX, NavPose) let your team start collecting data in minutes, not weeks. Supported hardware is recognized and configured automatically.
AI model management
Bring your own models. NEPI handles loading, inference, and output routing. Works with Darknet/YOLO and custom frameworks. Models connect directly to live sensor data, share outputs with downstream processes, and trigger automation actions.
Automation engine
Event-driven workflows. When this happens, do that. No custom code needed for common automation tasks and trigger responses. Turn AI detections into real-time actions: log data, send alerts, activate indicators, control actuators and robots.
Data collection
Structured, timestamped data from every sensor and process. Capture, synchronize, store, and export your data. Build high-quality datasets for AI model training and analysis without writing custom logging code.
Browser-based UI
Configure, monitor, and control your entire system from any browser. No apps to install. Works on any device with a web browser. Dashboards for system status, sensor feeds, AI outputs, and data management.
Built on ROS and ROS 2
NEPI is built on the most widely used frameworks in robotics and autonomous systems. The full codebase is publicly available at github.com/nepi-engine under the source-available license.
If your team is already working in ROS, NEPI adds the AI model management, hardware abstraction, and automation layer you would otherwise build yourself. No need to leave behind what you already know.
NEPI is not a robotics-only platform. It runs anywhere a team needs to connect sensors, run AI at the edge, automate tasks, and collect structured data.
What teams build with NEPI
From defense to education, across any environment where hardware meets AI.
Smart sensing
AI-enabled data collection at the point of capture. Process sensor data in real time, filter noise, and extract structured information at the edge.
Monitoring
Continuous environmental and system monitoring with automated alerting. Track conditions, detect anomalies, and log data without manual oversight.
Inspection
Automated visual and sensor-based inspection workflows. Detect defects, measure conditions, and generate reports from AI-driven analysis.
Automation
Event-driven control of actuators, motors, and robotic systems. Turn detections into actions in real time, from simple alerts to complex multi-step workflows.
Get started with the NEPI Docker Container
$ docker pull numurusnepi/nepi:latest
Resources
Everything you need to evaluate, install, and build with NEPI.
Documentation
Platform reference, API docs, and configuration guides for every NEPI subsystem.
Tutorials
Step-by-step guides from first install to building complete applications with NEPI.
Community
Join other NEPI developers on Discord and the community forum. Ask questions, share projects, and get help.
Ready to build?
Get NEPI running on your hardware in minutes. Free to try and source available, no sales call required.