Description
My client has a mobile application in development that aims to provide alternative credit-scoring to non-traditional borrowers in emerging markets - the unbanked'.
This app consists of 3 modules:
1. Psychometric questionnaire consisting of 40 questions in 4 groups of 10 that aim to measure psychological traits hypothesised to correlate with creditworthiness - for example honesty and financial literacy
2. Mobile data analytics to provide proxies for proof-of-address and other lending signals
3. Video analysis to provide other contextual clues relevant to the lending decision
They use a linear, rule-based algorithm to generate a score' from the user's answers to the psychometric questionnaire, and to indicate to a lender whether the user passed or failed the creditworthiness check.
We are looking for someone who can help us with the research and development on a part time contract basis, interviewing now with an eye to starting in the new year.
The project we have is ongoing research into extracting 'features' from 30 second video clips provided by our users - these could be things like facial expressions, gestures, background objects, etc. This could be finding and utilising 3rd party services and code, or by training our own computer vision models.
- Facial recognition/head-tracking
- Facial expression extraction from video
- Gesture recognition/extraction from video
- Gaze/eye-tracking from video
- Pose estimation from video
- Object recognition/extraction from video
This role is for ~20 hrs per week on a contract basis, remote, UK or nearshore candidates (close to GMT timezones). Happy to work asynchronously via Slack and Jira.