Very Short,Very Smart Resume
Hiring is a not a deterministic process. Job Descriptions are hard to define exactly. Its even harder to find exactly matching resumes. We decided to apply lean principles to deal with this situation
- Short Iterations, Quick feedback: Show some resumes – get some quick feedback – use the feedback to improve the next batch- keep the batch size small to keep getting quick feedback at the end of each iteration. This seemed counter-intuitive to the hiring managers and recruiters who believed that sending more resumes in bigger batches improved their chances of finding a match. But they have now started liking the idea because of the time saved.
- Scoring and ranking based on the Job Description: All requirements in the job description aren't equally important. Some are must have and others are good to have. Our algorithms using natural language processing carefully parse each word to assign weightages to each requirement. In parallel the software also compares how well the resumes in the pipeline satisfy the requirements and assigns scores to all the resumes. Initial ranking is based on these scores.
- Observe patterns, Use them to rank resumes: Hiring managers have specific likes and dislikes. History will show that certain attributes in a resume are liked more than certain others. Regular users get the advantage of our Machine learning algorithms that capture these patterns to rank the resumes .
- Transparency and feedback:We limit the batch size to top 3 resumes- thus enabling short iterations and quick feedback. We can't really demand feedback from the hiring managers if we don't tell them why a particular resume was considered. Feedback demanded is well justified because we share the scoring computation with the hiring managers and the candidates. Thus every one knows the reasons why the top 3 rankers are considered better than the rest.
- Learn from the feedback: We strive to improve our selection with each iteration by changing the selection criteria based on the feedback received.