Building Data pipeline on Edge devices such as Raspberry pie Real world Application

Hello, Atter Building A Data pipeline in my production Environment and applying Some transformations on my data using TFX I need to build the system pipeline for my serving model on my edge device.
I also want to monitor the incoming data to check if their a Concept or data drift and to detect any anomalies.
I’m trying to deploy my model on a Raspberry Pie since it is sufficient to run my model, and Professor Robert has advised in the course to replicate the same Data Transformations for the production and serving Environment. I am trying to install TFDV, TFX, and Apache beam
I started by checking compatible versions between Tensorflow, TFDV, TFX, Apache beam, and the Python required. after installing the appropriate Python as well as Tensorlfow, I’m not able to install TFDV or TFX on my Pie I’m not sure if it’s even available for Raspbian OS.

Why not use docker?

I Need specific versions of some libraries, If I use docker I would have to build the image myself

Isn’t it better to build an image than not being able to install on the target device?

Thanks for your response, However, what I need is confirmation from an expert that the technology is supported in an embedded device such as Raspberry Pie. And a general guideline or a hint to follow.

Post your question on

for expert advice or even better, hire a freelancer who might be working in the tfdv team.

Out of curiosity, why are you not building an image and trying it on raspberry pi?

If it’s not supported on Raspbian Os Docker isn’t an option.

As I suspected their is no release of TFX, or TFDV for Raspbianm, or even Arch Linux. as a general rule of thumb, Before building a complicated docker-compose image, check for compatibility on your target OS.

Adding @chris.favila

docker pull tensorflow/tfx

Thanks for TFX! if you find TFDV ARCH-based image please post it too.
Frankly, when I didn’t find TFDV ARCH Image I assumed that TFX also is not available.