This is a simple hiring app created to find top candidates in a dataset. Here is some of the key functionality:
- Scroll through the different candidates and get a high-level sense for their experiences:
- Deep-dive into a candidate's profile to learn more about their background:
- Filter and rank based on role:
These candidates are generated by creating the top 5 candidates for the role based on candidate rank. The candidate is ranked in three ways:
- A baseline rank accounting for various factors such as number of work experiences, highest education (biased most towards higher education), number of educational experiences (weighting more towards top 50 schools), and number of skills.
- A tags-based rank, comparing cosine similarity the candidates' job titles and educational subjects to pre-made keywords with an embedding model.
- An exact match to keywords, comparing exact matches between the candidates' job titles and educational subjects to pre-made keywords with an embedding model.
The keywords used in rankings 2 and 3 are the top 10 keywords for each role generated by GPT. The ranking is weighted most heavily towards rank 3, 2, and 1 respectively.