L&D Overview
Data labeling is an integral part of the data preparation process for machine learning and AI model development. The accuracy of an AI model is directly correlated to the quality of the labeled data used to train it. As a result, people with disabilities (PWDs), who work as data labelers for Vulcan, are the key contributors to determining the quality of label data and the precision of the AI model. To ensure employee performance efficiency and skills, the learning and development unit (L&D) has established an effective training and recruitment system.
The Vulcan PWDs' training system comprises of two main parts, which are the Reskill and the Upskill.
- Reskill training is a pre-employment process that aims to equip people with disabilities with the knowledge and digital skills necessary to work as AI trainers. The AI and Data 101 course is a fundamental course designed to help PWDs understand more about artificial intelligence, including
- 1.the evolution of AI: past, present, and future
- 2.AI in various industries, and
- 3.Data labeling for AI model training.
- Upskill training will be provided to PWDs after they get employed. The scope of upskilling courses ranges from data labeling training with an update of working principles (ground truth) to other essential skill sets required in this digital era. In addition, all lessons, materials, and the learning platform are specially designed with the consideration of accessibility for all disabilities users.
For the recruiting system of data labelers at Vulcan, we required PWDs to pass the reskill training courses of fundamental AI knowledge and basic digital literacy to be eligible to apply in the selection process. This strategic process will enable a pool of capable workforce. Moreover, throughout the training and application process, PWDs will receive assistance from local social development and human security offices in various regions that are in partnership with Vulcan. This recruitment system not only focuses on acquiring proficient workforces but also the efficient facilitating process to support future scalability.