Larry 黄昕

Quoting tool:
- Background: wholesale team didn't have a proper
quoting tool, in the old days we use excel version
rate cards. Quote was manually calculated through
the rate cards.
- What I contributed: an online web application with
built-in rate cards. It contains a front end (Vue) , a
backend (Python, flask) and a database (MySQL).
Service is built with Azure app service/storage
account.
- Output: Largely reduced the time to produce
quotes for general products.
Customer comments priority:
- Background: We have a platform where all the
communications during delivery are on it. The
platform doesn't have a priority/notification system to
notify project team even customer put comments on
it. Project team have to check orders one by one to
see the update from customer.
- What I contributed: A bot that consisting monitor
customer comments and categories them into
different kind/level or urgency and notify project
team. The core part is a sentiment-analysis model
(LSTM, PyTorch).
- Output: Project team now getting back to
customer faster and able to know the urgency
before browsing the orders.
Churn notification automation:
- Background: When services are ported, we
manually notify other operators. Step 1: gather
information from system logs. Step 2: validate those
information and send email to other operators. It is a
low value task consuming a lot of time.
- What I contributed: A information extraction( using
named entity recognition, nltk) that extracts
information and send out emails.