Smart List

website homepage
annoy model to predict the top five event matches
Brainstorming!

We are in Education Track

Inspiration

Oftentimes we as clients will receive spammy emails: they contain skills that we don’t match or areas out of our interests. In order to find our dream opportunities without tramp over a number of useless emails, we have decided SmartList, a website that matches our interests and skills to company jobs.

What it does and how we built it

SmartList bridges the need of companies and the interest / skills of clients. Upon request or weekly, SmartList will send clients emails that help them identify some of their featured dream jobs. This is done after client initially submits resume or LinkedIn profile, which generates, according to word frequencies in the resume, a number of feature vectors that describe their interests and skills, for example proficient languages and preferred working fields. A survey is then automatically generated to separate the interest vectors that are closest to each other. Furthermore, the clients can also modify their preferences from the automatically generated interests. Whenever companies release new activities or job posts, we will generate and compare the feature vectors and find the features that are the most similar to the seekers. We will sort the distance between feature vectors from wants and needs, with the least distance being the top fitted job. The top few jobs are released to clients in emails.

Challenges we ran into

Convert a pdf resume to txt file for counting the word frequency that help generating feature vectors for users.
HTML design.
Front and back end integration.
Selecting the correct database and integrate them into the algorithms.

Proudest accomplishment

Mapping out the entire project idea in one hour
Maximizing every teammate's expertise and potential
Keyword classification and frequency analysis using NLP model
Text extraction from LinkedIn profiles and sending personalized emails
Using annoy to figure out the top 5 opportunities
Having lots of fun and good food along the way!

What we have learned

Collaboration across diverse technical backgrounds, brainstorming, combining text extraction, nlp, knn models, web design all on one project

What's Next?

We really enjoyed the project: brainstorming together and learning from each other. Going into the future, we'd plan to bring this idea to life because we believe our project will improve the lives of many. Many ambitious and enthusiastic students will find the events they love on our platform. We will improve the UI design, integrate frontend with backend using Firebase, and make the algorithm faster from O(n^3log(n)) down to O(n^2log(n)) by implementing an algorithm for finding the shortest distance.