Transitioning from Python To C++ for EdgeAI/ML systems. Am I thinking about this right?

Context: I've built several mid/low-scale projects(around 200 users) like virtual prediction market(for competitve exam students with virtual coins) , Social forum(nextjs, gcp) with moderation pipeline in python, and have 3 more projects with python, JS, being the tech stack. I am starting my university in Taiwan (most probably) this year, and my focus is going to be on ML & Edge AI. I looked into tools like CUDA, TensorRT, etc and these seem to be most compatible with C++. I have previous experience with basic machine learning algorithms and libraries like sci-kit but all my work previous was done in python.

I am also getting into DSA and platforms like LeetCode and Codeforces(due to my interest), and I asked LLMs (gemini,chatgpt) whats the best way to go ahead and they suggested me to learn C++. Is this really the right way? Should I stick with python at start with DSA or get faimiliar with C++ and practise DSA in cpp?

TLDR: Fluent in python,js , Starting university interested in Edge AI & ML, confused if C++ is really the right way to go and if I should stick with python for DSA and learn C++ later (fear i wont be familiarised enough)

https://redd.it/1s7kp97
@r_cpp