InCommon Application

http://incommonapp.herokuapp.com

Purpose

Application designed to search for the common investment Portfolio Companies (PC) between two Venture Capital (VC) firms and view the details of each PC.

Demonstrate my understanding of the following technologies after my time as a fellow at Hackbright Academy:

Integrating an API
Using persistent & non-persistent storage to maximize performance
Web frameworks
Templating languages
Javascript, AJAX, and the jQuery library
Front-end frameworks

Stack

API: Crunchbase
Persistent storage: PostgreSQL
Non-persistent storage: Memcached
Backend: Python
Flask web microframework
Jinja2 templating for Python
Javascript (JQuery/AJAX) on the client
CSS based on Twitter Bootstrap

Features

Memcache - to enhance performance and rid of unnecessary API calls
Twitter's typeahead.js - provide search functionality for the 17,500+ VCs in Crunchbase's DB
Responsive design

Running

First, clone this directory to your computer.

$ git clone https://github.com/carlydacosta/finalprojectiqt.git

Create and activate a virtual environment in the same directory:

$ pip install virtualenv
$ virtualenv env
$ . env/bin/activate

Install the required packages using pip:

(env)$ pip install -r requirements.txt

Get a Crunchbase API key - https://developer.crunchbase.com/
Run the server
```
python incommonapp.py
```
Browse the application at [http://localhost:5000]

Breakdown

incommonapp.py: runs the program, contains the Flask routes
class_objects.py: contains the VC, PC, and Crunchbase classes
table_class_objects.py: contains the database table classes
incommon.db: store user, VC and PC information

Input VCs

View PC details

Experiential Learning:

The biggest pain point in my experience of building this application revolved around getting and storing the data. The challenges:

Exponential growth of API calls

API constraint of 50 calls per minute
My code blew through the constraint for just one VC and it's 50+ PCs

Time lag inherent in each call

Even if I called for the same info, it could take minutes before I got a response

Solution:

Memcache

This key-value store saves precious time in development and deployment
Non-persistent storage is great, but limits examining relationships.

Refactoring

Take user input of two VCs, make the API call for only their information
Make API call for a list of their PCs
Find the common PCs between them and make the API call for only those PCs
Make subsequent API calls for additional info relating only to them (image, funding round info)
Store the response for each call in memcache and the DB
One step before each API call however: first check memcache

Contact information

email: carly.a.dacosta@gmail.com

carlydacosta / incommon