📋 Loop 14
21/07/23
New updates and improvements.
Loop 14: releases, improvements and fixes
🎉 Releases
- Companies can now configure the application settings for each position to receive only the right type of applications, based on work eligibility.. Read more
- Update to work eligibility stream filters. Read more
- [internal] Data report to show number of new signups and reactivations from the past week.
📈 Improvements
- [internal] Changed from calling Google place API for CV location parsing to manually mapping Affinda formatted location object to a Google place object
- [internal] Installed eslint plugin to remove unused imports automatically
🔧 Fixes
- [internal] External check when bulk deleting external positions
- Fixed issue company user was experiencing with calendar integration
- Fixed bug where candidate with full work eligibility was not able to message positions
- Fixed bug with Lever integration
Hackathon
During this loop, we ran a cord hackathon. Please note, these features are not live on cord. The hackathon was an opportunity for cordians to experience the capabilities of AI, and how it could help our users.
Allowing users to build positions from a simple prompt: one team created a way for companies to create a cord position from a) an existing job description b) a career page URL and c) a custom prompt.
Build an AI that helps companies give detailed decline feedback to candidate applications: this team built a way for companies to give thoughtful feedback on a candidate’s profile based on both what matches and what doesn’t match.
Make position tagging automatic & improve mode for format positions and company profiles automatically: Instead of manually tagging required skills, this team built a way to automatically extract relevant information from job descriptions using AI. AI algorithms can identify and extract key skills or qualifications, making our internal tagging process faster and more accurate. Another part of the project was related to improving our current model for editing job descriptions and company profiles. This part of the project was built, and has been released, speeding up processes for the team that looks after external positions.
Build auto-verification: Using all of the data we have on past declines and acceptances, the team built a way to show in the admin console how likely a person is to be verified or declined, highlighting the fields that map to our criteria and flagging previous declines and fake profiles. … The first step to future auto-accepting and auto-declining!