Backend for UPC 2017
Advanced analytics used easily in a conversation
- Talk to your database like a human: We use advanced natural language processing techniques to create world class conversational experience
- Get data insights like you never did before: It’s never been so easy to identify factors that impact customers’ engagement and revenue.
- Smart AI-based database understanding: No complex setup required It's never been so easy to get advanced statistics
- Inside your favorite messaging app: One-click integration with Twist gives you more time to do stuff that matters.
TL;DR Stackshare
Our app consists of four main parts:
- Backend; Here we use microservice based approach with services build using Golang and Python.
- DevOps; We use AWS EC2 as our infrastructure provider. All our microservices are running using supervisor to provide demonization and easy restart/logging access. We use AWS Route53 for DNS related things. Every app is running behind nginx reverse proxy.
- AI; We use two frameworks. CatBoost for Python implementing gradient boosting on decision trees with a native support of categorical features and GoML for Golang to perform sentiment analysis via Naive Bayes classification.
- Frontend; We have build responsive landing page using ES6/Bootstrap/jQuery
returns all possible commands with descriptions
returns a picture that vizualizes the given query
returns data that mathches the given query
Use deploy.sh script in the root directory. You should have private key called key.pem in your .ssh directory and following line in your ssh config:
Host upc
HostName ec2-18-194-130-70.eu-central-1.compute.amazonaws.com
User ec2-user