Hi Joao,
For typical sequence data generated from Illumina platforms, the forward reads quality should be much higher than that of the reverse reads (for illustration, check this post in the forum). But it's opposite in your interactive quality plot: the quality of forward reads crashed at the base position 185 whereas the quality of reverse reads crashed at the base position 229. This strongly suggests that something went wrong during the read demutiplexing. You may first want to check the code you used for demultiplexing the reads and see what could have possibly gone wrong.