GPT-4o
⚙️ Technical
Intermediate
IoT Edge AI Data Processor
Design an on-device AI processing pipeline for IoT sensor data that runs inference at the edge with minimal latency.
The Prompt
# IoT Edge AI Data Processor You are an edge AI engineer. Design an on-device inference pipeline for IoT sensor data. ## Sensor Context - Sensor type: [SENSOR TYPE, e.g., "vibration sensor", "thermal camera", "industrial pressure gauge"] - Data format: [RAW FORMAT, e.g., "time-series CSV", "JPEG frames", "binary stream"] - Sampling rate: [SAMPLES PER SECOND OR MINUTE] - Edge device: [HARDWARE, e.g., "Raspberry Pi 5", "NVIDIA Jetson Nano", "STM32 microcontroller"] - AI task: [CLASSIFICATION / ANOMALY DETECTION / PREDICTION / OTHER] - Connectivity: [ALWAYS CONNECTED / INTERMITTENT / FULLY OFFLINE] ## Pipeline Design ### Model Architecture Recommend a lightweight model type (MobileNet, TinyML, ONNX-quantized transformer, custom CNN) suitable for [SENSOR TYPE] and [EDGE DEVICE]. ### Preprocessing Describe the data transformation steps before inference (normalization, windowing, feature extraction). ### Inference Engine Choose the runtime (TFLite, ONNX Runtime, TensorRT Nano) and explain why it fits [EDGE DEVICE]. ### Output Handling How inference results are stored locally, flagged for alert, or synced to a central system when connectivity allows. ### Power & Memory Budget Estimated RAM usage, inference time per sample, and battery impact if device is battery-powered.
📝 Fill in the blanks
Replace these placeholders with your own content:
[SENSOR TYPE, e.g., "vibration sensor", "thermal camera", "industrial pressure gauge"]
[RAW FORMAT, e.g., "time-series CSV", "JPEG frames", "binary stream"]
[SAMPLES PER SECOND OR MINUTE]
[HARDWARE, e.g., "Raspberry Pi 5", "NVIDIA Jetson Nano", "STM32 microcontroller"]
[CLASSIFICATION / ANOMALY DETECTION / PREDICTION / OTHER]
[ALWAYS CONNECTED / INTERMITTENT / FULLY OFFLINE]
[SENSOR TYPE]
[EDGE DEVICE]
How to use this prompt
1
Copy the prompt
Click "Copy Prompt" above to copy the full prompt text to your clipboard.
2
Replace the placeholders
Swap out anything in [BRACKETS] with your specific details.
3
Paste into GPT-4o
Open your preferred AI assistant and paste the prompt to get started.