proposed workflow for loading surface geometry per https://github.com/shapely/shapely/discussions/1369

surfaces.py

from geomet import wkt
from io import StringIO
import psycopg2


def main():
    # we'll use this to copy data into postgres
    fileobj = StringIO()

    # build WKT geometries
    for idx, surface in enumerate(my_surfaces, start=1):
        geojson = {
            'type': 'polyhedralsurface',
            'coordinates': [[
                [triangle] for triangle in surface
            ]]
        }
        # this will not work yet - would require work from geomet library
        ewkt = f"SRID=4326;{wkt.dumps(geojson)}"
        fileobj.write(f"{idx},{ewkt}\n")

    # seek back to the beginning of the file object
    fileobj.seek(0)

    sql = """
    create table surfaces_tmp(
        id int,
        surface geometry(PolyhedralSurface, 4326)
    );
    """
    with psycopg2.connect(**params) as conn:
        cursor = conn.cursor()
        cursor.execute(sql)
        # copy will load lots of data into postgres very quickly.
        cursor.copy_expert('copy surfaces_tmp from stdin with csv', fileobj)
        # you can also perform any other required post-processing here with postgis functions.
        cursor.execute('insert into <my table>(surface_geometry_column) select surface from surfaces_tmp')