What's that diode doing in the GND lead of the Nano? Depending on the diode you use, it will lift the Nano's GND by 0.25-0.75V vs. the motor module. Same story with the diode in the 3.7>7.5V buck-boost converter - what's it doing there? These things only complicate matters.
Btw, why not just use a 3.7 > 5V buck-boost and bypass the Nano's linear regulator? It's only dissipating power as it is. Not much, but why have use it anyway if it's not necessary.
Where are the connections between the Nano, the level shifter and the MPU6050? I only see what seems to be GND.
Anyway, as to your problem, there's a likelihood your juggling with diodes is causing problems or at least contributing to it. Perhaps you've already included the cap between the Vin on the Nano and its (lifted) GND, but what value is it? If it's smaller than, say, 10uF, try something like 100uF or bigger.
If all else fails, you could include a small RC circuit on the Reset pin. Since a 1k (possibly depending on the Nano version you use) pullup is already present, you might try something like 1uF parallel to the Nano's reset button. This will hold Reset low for a bit as the power supply stabilizes. DC-DC converters do sometimes exhibit startup behavior that microcontroller's don't like.