The modern internship search is flawed. It’s filled with scams that collect data, fake job listings, and dishonest recruiters. Vera is not just a simple interface for a proprietary API or a basic text classifier using a fixed dataset. It is a privacy-focused, locally hosted AI tool designed to restore honesty to the job market.
To enable real-time detection of scams without the delays, costs, or privacy concerns of cloud-based APIs, Vera avoids traditional large-scale systems. Instead, it uses a highly efficient SLM+LLM hybrid pipeline. Fast, lightweight models that use DistilBERT and an LSTM network manage rapid sequence classification. They parse messy DOM elements and informal social media feeds to immediately identify known language issues and phishing signs. Only complicated or unclear postings go to a localized LLM for a thorough examination of recruiter intent and company legitimacy.
Moreover, Vera functions as a multi-agent system. While the hybrid pipeline takes care of natural language processing, specialized agents work independently to gather information from various job boards, check public registries, and analyze past hiring patterns. Presented through a responsive and user-friendly interface, Vera easily integrates its insights into the student’s workflow. By merging localized machine learning with widespread fact-checking, Vera provides a clear, open-source defense against job fraud.