interests in 17. Optimized Serving endpoint for classic machine learning model training and inference / questions

[A7med7x7]

Hello, My name is Ahmed Alghali, Computer Science Student at the University of Khartoum, Sudan, I have actively applied my skills in different Machine Learning Scenarios from model development to deployment, achieving a top 5 ranking on the Zindi platform(Machine Learning Competition) in the Sudan region and placing within the top 1% of users. My classical machine learning background and mathematical foundations helped me win multiple honors and regional prizes within the computational field.

I have been exploring the Optimized Serving endpoint for classic machine learning model training and inference project
within your organization, I find it as a middle ground between what I know and want to learn.

What I’m Familiar with (what I want to hone)
Python programming, machine learning, Apache airflow, and setting jobs.
What I’m not familiar with (What I want to learn under your mentoring),
onedal
API's end points

Could you advise on whether this project might suit me or if it's too much to learn and grasp while being part of it? I got interested in the idea, but it's considered new to me.

[mzegla]

Hi Ahmed, sorry for the late response. Thank you for your interest in this project.

It basically consists of two parts:

• preparing Python logic that will be responsible for model training (here you would use Python and packages like already mentioned onedal)
• embedding this logic into the model server (it already has so called Python nodes - https://docs.openvino.ai/nightly/model-server/ovms_docs_python_support_reference.html - that could serve such python logic)

Model Server core is written in C++, so you will likely have to spend some time looking into it, especially the parts responsible for API <-> Python Node flow in order to synchronize all parts of your work.

So looking at your background you will have a part that you are familiar with (Python code, ML algorithms etc.), but also C++ parts related to serving aspects which might be new for you.