Matchbox

Matchbox is a quick and easy service that will pair people with companies based on their skillsets to help them find the right fit.

Comment

Inspiration

We are all graduating college soon and are on the job hunt! We thought we could make the hunt a little easier by helping people find companies that might be a good fit for them based on their skills and qualifications. Also, we hope to give a big picture view of companies that people might be interested in by giving them insight into popular sentiment and valued skills.

What it does

Matchbox is a service that will match you based on your resume or LinkedIn profile to companies that align closest to the skills and qualifications that you possess. With data sourced from Glassdoor and LinkedIn, it also shows data visualizations of metrics such as popular sentiment over time, common words used to describe the company, and more.

How we built it

This project is inherently data driven, so gathering the data was a fundamental part of our development process. We built web scrapers to pull reviews from Glassdoor and also pull company information from LinkedIn. Additionally, we performed sentiment analysis on the reviews from Glassdoor to see what popular opinions of a given company would look like. On the frontend, we have beautiful data visualizations powered by dc.js, which allows for dynamic crossfiltering so you can clearly see relationships between all of our data.

Multi-region Fault Tolerance with CockroachDB

In line with our focus on data, we wanted to take the challenge of coming up with a fault tolerant and portable deployment scheme for our app. Using CockroachDB's Helm operator, we were able to deploy a multi-region fault tolerant CockroachDB instance on 3-node Kubernetes clusters running on both the east and west coast on Linode and GCP respectively. Along with load balancing between the clusters, this allowed our service to survive an entire data center failure in one of the availability zones. Getting this working under a tight Hackathon timeline was a challenge, but the more rigid data model and consistency guarantees that CockroachDB afforded us were worth the time investment. In addition, we gained a lot of valuable knowledge about Docker, Kubernetes, and networking between cloud providers.

Challenges we ran into

This project was an ambitious undertaking, so we did have to table some ideas we had for the future. One lesson we took away from this project is that if we had access to the APIs for Glassdoor and LinkedIn, it would be much easier to get at the data that we need instead of having to roll out our own scrapers and grab at the data ourselves, which is a lot more resource-heavy and time-consuming.

Accomplishments that we're proud of

At the end of the day, we're proud of the data we were able to gather and along with the visualizations we created, we believe that this could be genuinely valuable to a lot of people, especially new grads, seeking employment.

What we learned

A lot of the visualization was relatively new and challenging for some of us, so it was a great experience to get some exposure to an easily accessible and powerful tool. Also, sentiment analysis was a topic that a few of us wanted to explore at some point, so we're glad we had the opportunity to incorporate it in our project.

What's next for Matchbox

As it stands, there's a lot of room for optimization when it comes to the gathering of data. If we can get authorized to use Glassdoor's API or even get accepted into the LinkedIn Partner Program so we can have full access to all of their API endpoints, it would streamline the data collection process for us. In the future, we also hope to help prospective employees grow their network by matching them with people working at the companies that they think would be a good fit for them. All in all, there is a lot of different directions that could be taken with this project, and we're confident that it can help people on the job hunt.

Built With

Share this project:

Updates