[pacnw]

First, do know that the scope you have outlined is a huuuge undertaking. Maybe think of limiting the scope to start off with (inventory mgmt OR dealer mgmt OR accounting OR payroll). For tech stack, my suggestion is to use a back-end that provides as much out of the box as possible.

Ruby on Rails provides huge bang for the buck with their integrated database ORM, as does Phoenix (what I used). Your scope involves a lot of referential data, stick to traditional robust SQL database system (Postgres for example). If you use a NoSQL db, you may be quicker out the gate (as it'll take any JSON you give it), but once you get to the analytics with lots of joins, it won't be as much fun (personal experience). Also, I would suggest using the server frameworks' rendering engine to get your MVP up and running, as you can then fast iterate between data model changes and visual representation (i.e. HTML output), and avoid front-end builds and all that encompasses. Phoenix is very snappy, and with something like Turbolinks can potentially be as snappy as a dedicated front-end app.

Personally, I needed very interactive front-end, so had to build a separate front-end using Ember. I found Ember with Ember Data provides a robust toolset that manages complex data. I also have experience with React and Vue, but keep in mind with their state management (e.g. Redux / Vuex) that you'll need to manage complex data relationships yourself. As far as mobile, don't think native, as you'll need to learn specific technologies to implement that as well (more time...). Grab a bootstrap template that is already responsive, and you can run the same web-site on an iPad (forget the phone initially). Good luck!

source: just finished a 2 year build of an investment-related accounting and analytics webapp