Get Me Job

Inspiration

We write our resume, but we are not aware of the job positions out there. 'Get Me Job' simplifies the search by showing job openings that you are eligible and suitable for.

What it does

Given a resume, it finds roles that you are qualified for and aligned with your experience.

How we built it

We crawled 1200 job descriptions on Lever.
We parse the job description with FireFunction function calling.
We indexed the result on Vectara.

When the user uploads their resume

We use LlamaParse to parse the resume.
We use GPT-4 function calling to parse the resume.
We use fields such as years of experience and education attainment to filter job description.
We use semantic matching to rank the most relevant job descriptions for the resume.

Note: We did not manage to get Vectara working at submission time.

Challenges we ran into

There is currently no easily available API that we can call to query for the jobs available. We bootstrapped this by crawling AI companies with jobs on Lever.

Accomplishments that we're proud of

We managed to use function calling to parse the resume and job description.

What we learned

Function calling (FireFunction, GPT-4) is currently already quite reliable at producing usable JSON which we can use to index and query job descriptions. However, if we require nested outputs (for example the list of experiences, we still need to use OpenAI GPT-4).

What's next for 'Get Me Job'

More prompt engineering to implement more precise definitions Index a greater amount of job positions Add AI career advice for resume and interview prep. Launch a mobile app for on-the-go job searches.