He mostly talks about practical engineering issues: modular reactor construction, easy maintenance, testing current flow through joints in superconducting tapes, 3D printing a reactor chamber with complex cooling channels, immersion in FLiBe for ample cooling and breeding 14% more tritium than the reactor consumes, etc.
The reactor would be about the size of JET, which was built in four years. MIT has pretty good experience for this, since their Alcator C-Mod has more powerful magnetic fields than any other tokamak in the world.
Of course the startup companies pursuing fusion are also taking an engineering approach. Here's an interesting panel discussion[2] between the MIT guy and people from Tri Alpha, General Fusion, and Lockheed.
I would be the first person to tell you that I am way further on the D side of the R&D spectrum than the R side.
What does that say for software?
As an engineer I much prefer the latter, something with a bit of urgency that "we need to develop this as quickly as possible because it unlocks a lot of solutions to problems which are threatening billions of people." Not "Ohh look at all the shiny ways we can turn three two small atoms into one slightly bigger atom!"