Artificial intelligence defeated the system of personnel selection in a technology company.
A user of the social network X under the nickname facelessboy00 shared his unusual way to help a friend get a high-paying job in a technology company. As an experienced developer, he used deepfake technology to replace his face with that of a friend during video conversations.
According to the developer, his friend took a course in frontend development, but because of a lack of self-presentation skills, he could not find a job for three months. Realizing that his friend was in danger of returning "to the factory", he decided to help him.
The first step was to create a "selling" resume with work experience in little-known companies. To do this, LinkedIn selected profiles of people from the domain domain who work in large companies. Items that would look good were borrowed from the selected profiles. Then they were transferred to the final resume, and as experience they added work experience in little-known companies lasting up to 2.5 years.
Then we started preparing for the interview using the DeepFaceLive program, which captures the video and returns it with the already substituted face. The delay was about 500 ms. A dataset of 10,000 photos of a friend was used to train the neural network. The training process took 14 hours on an RTX 3070 graphics card, which later had to be upgraded to RTX 4090.
At the first interview, they were found out because at some point the mask almost "flew off". Facelessboy00 was too relaxed and turned away, breaking the synchronization. After that, they trained the model again for a week to improve the quality.
At the second interview, everything went well, and the friend was offered a job with a salary of $ 4,000 per month.
Now he is preparing for onboarding in the company, and facelessboy00 promises to tell about the further results of this unusual experiment.
A user of the social network X under the nickname facelessboy00 shared his unusual way to help a friend get a high-paying job in a technology company. As an experienced developer, he used deepfake technology to replace his face with that of a friend during video conversations.
According to the developer, his friend took a course in frontend development, but because of a lack of self-presentation skills, he could not find a job for three months. Realizing that his friend was in danger of returning "to the factory", he decided to help him.
The first step was to create a "selling" resume with work experience in little-known companies. To do this, LinkedIn selected profiles of people from the domain domain who work in large companies. Items that would look good were borrowed from the selected profiles. Then they were transferred to the final resume, and as experience they added work experience in little-known companies lasting up to 2.5 years.
Then we started preparing for the interview using the DeepFaceLive program, which captures the video and returns it with the already substituted face. The delay was about 500 ms. A dataset of 10,000 photos of a friend was used to train the neural network. The training process took 14 hours on an RTX 3070 graphics card, which later had to be upgraded to RTX 4090.
At the first interview, they were found out because at some point the mask almost "flew off". Facelessboy00 was too relaxed and turned away, breaking the synchronization. After that, they trained the model again for a week to improve the quality.
At the second interview, everything went well, and the friend was offered a job with a salary of $ 4,000 per month.
Now he is preparing for onboarding in the company, and facelessboy00 promises to tell about the further results of this unusual experiment.
