The AI Engineer Roadmap should have Model Optimization as a topic and its concrete resources

[lalchandaniaadarsh]

Roadmap URL

https://roadmap.sh/ai-engineer

Suggestions

It should have model optimization techniques and model compression techniques to deploy a neural network on production to reduce latency and deployment costs at the cost of some accuracy.

[github-actions]

🙌 Hello! Thank you for taking the time to file an issue.

If this is a bug report, please include any relevant logs or details that can help us debug the problem. Your help is greatly appreciated! 💡

We'll get back to you as soon as possible, kindly be patient for a response from a maintainer.

[antoninoLorenzo]

deploy a neural network on production

The AI Engineer roadmap doesn't really cover a lot of the basic topics you need to understand how to "deploy" a "neural network" in "production", there are various topics involved such as:

• deep learning basics (-> math), and this is covered in AI and Data Scientist,
• evaluation both offline and in production, that is partially covered by MLOps (monitoring and observability)
• some backend development knowledge is also involved when you need to build production-ready stuff, and that is covered in API Design and Backend Developer

Given that the AI Engineer roadmap has a more top-to-bottom approach I think that either it gets redesigned to include the required knowledge in the first place, or just a few topics are added under a Operationalize (or something similar) card; the most important topics I can think of are more related to LLM evaluation/monitoring rather than optimization (you should know what to optimize first).