Frustration with getting unsolicited job offers based on text-processing algorithms, which, in the end, still did not fit our candidate profile.
What it does
We used the InfoJobs API to gather approximately 49,000 open job offers, and trim them down to the most relevant details for a jobseeker.
We then found a data-set of ~1200 CVs online and began to work towards implementing Natural Language Processing across all of these CVs and job offers in order to form clusters to identify similar jobs and the CVs that match them the best.
Additionally we later worked on adding in a web interface to throw this data back to users that would be able to upload their CV, and be returned different jobs in different categories that fit their profile.
How we built it
Challenges we ran into
We ran into a challenge trying to automatically translate all the InfoJobs dataset that we've pulled through the API. The raw size of the dataset made it hard to use automated cloud tooling without getting rate-limited.
Cleaning the dataset took quite a while.