Approach reviewing an existing codebase - AI Code

How do you usually approach reviewing an existing codebase to understand exactly how to continue working on it and where to pick things up, especially when the project was initially built with AI? I'd say it's a -large integration project rather than a simple application. synchronization system with multiple modules and interconnected services, involving Tally XML communication, Zoho APIs and webhooks, database operations, authentication, and synchronization workflows.

I haven't measured the codebase in terms of LOC or file count yet. My main challenge is getting a clear picture of the overall architecture and understanding how the existing services interact before starting to add new ones.

For example, imagine a project with too many.md files containing instructions for AI agents, explaining how they should work from scratch. Now you're asked to start adding new services to the project.

What's your usual approach in this situation? How do you get familiar with the project and understand its architecture and existing implementation? Do you actually go through all those instruction files, or do you follow a different approach?

I'd love to hear about your experiences and how you typically handle this kind of onboarding. sine I am a junior:)