Python/Edge-AI Developer:Edge Impulse Vision Model to GPIO/PWM Hardware Trigger (RaspberryPi/Jetson)

We are building a wet-environment AI prototype for the agricultural tech sector (specifically, a system that detects and targets poultry pinfeathers using computer vision and a high-power laser). I am looking for an experienced Python / Embedded Systems Developer to write the low-latency "glue code" that connects our AI vision pipeline to the physical hardware. The Hardware Architecture (Already Engineered & Sourced): I have already fully engineered and solved the physical, optical, and thermal challenges of this build. You will not be responsible for designing the hardware or enclosure. The physical system consists of: Compute: Nvidia Jetson Orin Nano Vision: Arducam USB Camera Effector: 10W (Optical) 450nm Blue Laser Module (TTL/PWM controlled) Housing: IP68 Hammond Diecast Aluminum heatsink enclosure with 1.5mm uncoated Synthetic Sapphire optical windows to prevent thermal shock in a wet environment. Power: Fully isolated 19V (Jetson) and 12V 60W (Laser) power delivery. Your Scope of Work: We need you to write the robust, precise Python code to tie these components together: Ingest the live Arducam video feed (via OpenCV/GStreamer). Interface with our computer vision model to track target coordinates. Translate those targeting events into a precise 3.3V TTL/PWM signal sent via the Jetson’s GPIO pins to fire the 10W laser with millisecond accuracy. Ideal Qualifications: Strong experience with Python and the Nvidia Jetson ecosystem (JetPack, Jetson.GPIO). Experience with OpenCV and minimizing latency in camera streams. Proven experience interfacing software with physical hardware (triggering PWM/TTL signals, controlling CNC/lasers/actuators). An understanding of fail-safes (we are working with a Class 4 laser, so software-level fail-safes to prevent the laser from firing indefinitely are a must). If you are an expert at bridging the gap between AI and physical hardware, please apply. We have the Bill of Materials finalized and are moving into the physical assembly phase immediately. Please start your proposal with the word "SAPPHIRE" so I know you have read the hardware specifications. Urgent (Needed within 3-5 days) The Hardware Stack (On my desk): Compute: Nvidia Jetson Orin Nano (or Raspberry Pi 5 with Coral TPU) *[Note to applicant: Please state which you prefer to write for]* Camera: Arducam Global Shutter USB/MIPI camera Actuator: A hardware module controlled via a 5V/3.3V PWM or GPIO logic pin. Project Milestones & Deliverables Deliverable 1: Camera Ingestion & Low-Latency Pipeline A Python module that successfully accesses the Arducam USB camera via OpenCV or GStreamer on the Nvidia Jetson Orin Nano. Must achieve a stable, low-latency video stream (minimum 30 FPS) without dropping frames or overloading the Jetson’s CPU. Deliverable 2: Laser Control & Safety Module (Jetson.GPIO) A Python module dedicated to controlling the 10W laser via the Jetson’s GPIO pins. Must be able to receive a firing command and send a precise 3.3V TTL/PWM high signal for a specified millisecond duration. Crucial Safety Fail-safes required in this code: A hard-coded maximum firing duration (e.g., laser cannot fire for more than 500ms continuously to prevent fires/hardware damage). A software "kill switch" routine that instantly forces the GPIO pin to LOW (0V) if the program crashes or the camera feed drops. Deliverable 3: Camera-to-Laser Calibration Script (Addressing Parallax) Because the camera and the laser will be looking out of two separate 30mm Sapphire windows side-by-side, there will be a physical offset (parallax). You must provide a simple calibration script/GUI that allows me to click on the screen where the laser dot physically lands, generating a coordinate offset matrix so the AI knows exactly where the laser is pointing relative to the camera's field of view. Deliverable 4: The Core "Glue Code" (Main System Loop) The final, multi-threaded (or asynchronous) main Python application that ties everything together. The Loop: Ingests frame - Passes frame to our AI model - AI identifies target coordinates - Applies calibration offset - Triggers GPIO laser module for X milliseconds - Logs the event - Repeats. Must run efficiently on the Jetson Orin Nano's architecture without blocking the video feed while the laser is firing. Deliverable 5: Deployment & Handoff Documentation A clear README.md or requirements.txt detailing the exact Python environment setup, dependencies, and JetPack configurations needed to make the code run natively on my Jetson motherboard. Instructions on how to configure the script to run automatically on boot (Systemd service) for headless field testing.