SnifTern.ai: A Machine Learning-Powered Platform to Protect Job Seekers from Internship Scams

SnifTern.ai is an open-source project based on Flask and machine learning, focusing on identifying fake internship recruitment information and verifying company authenticity. It provides an intelligent defense line for students, addresses the rampant issue of internship scams, and offers a reliable risk assessment tool for students, educational institutions, and employers.

Internship scams are a global problem. Scammers exploit students' career aspirations to发布虚假信息 and defraud money or personal information; traditional manual review is inefficient and prone to omissions. SnifTern.ai aims to use automated machine learning technology to analyze recruitment information in real time, identify fraud signals, and provide risk assessment tools.

The backend uses Python Flask, and the core process includes four stages: data ingestion, preprocessing, analysis and scoring, and report generation. Core features:

1. Real-time fraud analysis: Evaluate the wording of job descriptions, salary rationality, and completeness of company information;
2. Cross-platform integration: Cross-verify information consistency from LinkedIn/Indeed/Glassdoor;
3. Company legitimacy verification: Check enterprise data points, digital footprints, and public records to estimate credibility.

Hybrid detection strategy: Rule engine + machine learning model. NLP analyzes text features to identify scam话术; anomaly detection finds data points that deviate from normal job positions; supervised learning predicts fraud probability based on historical data. Detection results include risk scores, confidence levels, and explanatory labels, allowing users to understand the basis for judgments.

Practical features:

1. PDF report generation: One-click generation of professional reports containing risk scores, evidence, and recommended actions, which can be shared with mentors/career centers;
2. Multilingual support: Built-in language detection and translation to serve global users;
3. Modular architecture: Microservice design allows replacing models/data sources, and the rule engine can adjust detection rules without redeployment.

Privacy and ethics: Only ingest public recruitment information, respect platform terms and user consent; data processing follows access control, audit logs, and minimization principles; risk scores are accompanied by explanations and evidence tracking to support auditability; adhere to ethical guidelines for data use and user safety.

Deployment: Lightweight and easy to run, uses SQLite for development, can be migrated to PostgreSQL/MySQL for production, and supports Docker containerization; RESTful API can be integrated into systems like school portals. Scenarios: Educational institutions batch review positions, individuals verify before job hunting, recruitment platforms improve content quality. MIT open-source, welcome community contributions such as model improvements and data source expansion.

Future plans: Expand more recruitment platforms/regional data sources, improve model accuracy through user feedback, add language packs and visual reports, and build a plugin market for easy integration. Conclusion: SnifTern.ai uses AI to automate scam identification, reduce students' job-seeking risks, demonstrate the potential of technology to solve social problems, and such tools will become more important as online recruitment becomes popular.