Blocked updates and background writer performance

PG 9.4.4 (RDS)

I'm experiencing an issue when trying to update many rows in a single table (one row at a time, but parallelized across ~12 connections).  The issue we see is that the writes will periodically be blocked for a duration of several minutes and then pick back up.  After digging through our monitoring stack, I was able to uncover these stats which seem to allude to it being a background writer performance problem:

(apologies for the image)

Our settings for the background writer are pretty standard OOB (I threw in some others that I thought might be helpful, too):

          name           | setting | unit

-------------------------+---------+------

 bgwriter_delay          | 200     | ms

 bgwriter_lru_maxpages   | 100     |

 bgwriter_lru_multiplier | 2       |

 maintenance_work_mem    | 65536   | kB

 max_worker_processes    | 8       |

 work_mem                | 32768   | kB

The table that is being written to contains a jsonb column with a GIN index:

It seems to me that the background writer just can't keep up with the amount of writes that I am trying to do and freezes all the updates. What are my options to improve the background writer performance here?

thanks

--Cory