# Develop applications

Use the Dragonwing IQ-9075 EVK to create AI/ML and robotics applications. The following developer workflows explain the process for building, integrating, testing, and deploying these applications.

## AI/ML developer workflow

The AI/ML developer workflow begins with optimizing models using tools like AI Model Efficiency Toolkit (AIMET) and converting them to formats such as Lite Runtime (LiteRT) or ONNX. These models are then integrated into applications using SDKs like the Qualcomm^®^ AI Engine Direct or Qualcomm^®^ Intelligent Multimedia SDK (IM SDK). Runtime configurations and hardware compatibility are addressed during integration.
The applications are deployed to edge devices, leveraging accelerators like DSPs or NPUs for efficient inference. You can validate performance and accuracy under real-world conditions.

Using the [AI developer workflow for Ubuntu](https://docs.qualcomm.com/bundle/publicresource/topics/80-90441-15/introduction.html), you can do the following:

> 
> 
> 1. Optimize models using tools such as AIMET.
> 2. Convert the models to formats such as LiteRT or ONNX.
> 3. Integrate the models into applications using SDKs such as Qualcomm AI Engine Direct or Qualcomm IM SDK.
> 4. Configure runtime settings and ensure hardware compatibility during integration.
> 5. Deploy applications to edge devices, using accelerators like DSPs or NPUs for efficient inference.
> 6. Validate performance and accuracy under real-world conditions.

## Robotics developer workflow

The robotics developer workflow starts with setting up the environment using the Qualcomm Robotics SDK, including ROS 2 and GStreamer. You can build and flash the OS onto platforms on the device to prepare for application development. You can also prototype features such as object detection, VSLAM, and sensor fusion using SDK tools.
The modular SDK architecture and standard APIs support scalable and reusable development. Applications are deployed across various robotics domains, ensuring adaptability and performance.

Using the [Robotics developer workflow for Ubuntu](https://docs.qualcomm.com/doc/80-90441-2/topic/1-introduction.html), you can do the following:

> 
> 
> 1. Set up the development environment using the Qualcomm Robotics SDK including ROS 2 and GStreamer.
> 2. Build and flash the operating system on the device to prepare for application development.
> 3. Prototype features like object detection, VSLAM, and sensor fusion using the SDK tools.
> 4. Use the modular SDK architecture and standard APIs, which support scalable and reusable development.
> 5. Deploy the applications across various robotics domains, ensuring adaptability and performance.

Last Published: Apr 17, 2026

[Previous Topic
Run sample applications using Qdemo](https://docs.qualcomm.com/bundle/publicresource/80-90441-252/topics/run-sample-apps.md) [Next Topic
Development device regulatory notice](https://docs.qualcomm.com/bundle/publicresource/80-90441-252/topics/Development_regulatory_notice_5.md)