Amanda and Jeff talk about their latest project: implementing the microservice layer of KillRVideo in Python. They talk about what was easy, what was not too easy, and what was just plain fun. They spend some time on discussing recommendations engines and ask the audience should they implement this in DSE graph or using PySpark or other python package? Comment below!
0:24: Jeff and Amanda talk about how and why they wanted to add a Python implementation of KillRVideo
1:06 Jeff asks if apps/services are actually written for production with Python
1:15 Amanda talks about how Python is one of the top languages and that developers do develop for production in Python.
02:08: Jeff mentions Python and Data Science and rapid development
03:14 Jeff and Amanda share a Little know fact that cqlsh is written in python and uses the python driver
03:37 Only reimplementing on the microservices in Python (not the front in)
04:10 Jeff: What was the easiest part of this project
04:30 Amanda: It was important to Amanda that there was a lower barrier to entry for KillRVideo and Python provides that
05:05 Jeff: He increased his productivity by 3x by prototyping in the console.
05:50 Amanda: Loves python because its so easy to learn and get things moving up and going. All the libraries help with that.
06:23 Jeff: No need to use REGEX, can just import a library to do email valid form checking
6:49 Amanda: Asks Jeff about how it was to learn how to use the Python driver. --It was easy!
07:44 Jeff: The least easy part was using the mapper
07:58 Amanda: When to not use the mapper?
08:23 Jeff: Can't use paging with mapper need to manage that yourself
08:50 Jeff: So easy to use Search queries and was so easy that it just worked
09:24 Jeff: What was not so easy?
09:40 Amanda: Dependency hell!
10:18 Amanda: Use virtual environments (Anaconda etc)
11:00 Jeff: Did not like the lack of strong typing in Python. Typing doesn't resolve until runtime.
11:44 Jeff: Java is his jam. Java for the win. Strong typing all the way.
11:55 Amanda: We talked about what was easy, what was not so easy, but let's talk about what has been fun!
12:45 Jeff: Fun to learn how to use Kafka!
13:29 Jeff: Using graph for our recommendations.
13:51 Amanda: We can use graph, but we can use python for data science
14:10 Amanda: Amanda get stumped -- the answer is Collaborative Filtering
14:30 Amanda: Okay audience how should we implement our recommendation engine? Graph OR python?