Hi @Bruno Franco​ ,

Can you please try the below code, hope it might for you.

from pyspark.sql.functions import from_unixtime
from pyspark.sql import functions as F
final_df = df_src.withColumn("Final_Timestamp", from_unixtime((F.regexp_extract(col("CreateDate"), "(\d+)", 1)/1000),"yyyy-MM-dd HH:mm:ss"))\
                 .withColumn("Offset_Time", F.regexp_extract(split(df_source['CreateDate'],'\\+')[1], "(\d+)", 1))
display(final_df)

I have divided the value with 1000 because from_unixtime  takes arguments in seconds, and your

timestamp is in milliseconds.

Happy Learning!!