How kandagatlaabhijith/veriintern-2026 works
VeriIntern is a web-based internship-posting fraud analyzer. A user submits posting text (and optionally a URL) through the React interface; the FastAPI analysis endpoint combines model, rule/feature, and URL signals into a risk result. Separate endpoints inspect URLs and payment language and explain predictions. The backend also provides dataset-based model training and benchmark evaluation. Some frontend request details and backend fusion logic fall in omitted excerpts, so those relationships are marked unverified rather than asserted as confirmed calls.
Main entry points: main.py FastAPI endpoints · part1_dataset_similarity.py Feature extraction · benchmark_eval.py Benchmark evaluation · part2_ml_model.py ML ensemble
Components 12
Web client
- Analyzer interface
frontend/src/App.jsx
API orchestration
- FastAPI endpoints
backend/main.py
Detection signals
- Feature extraction
backend/part1_dataset_similarity.py - Text normalization
backend/normalization.py - Payment detection
backend/payment_detector.py - URL risk analysis
backend/url_checker.py - Domain trust
backend/trust_layer.py - Prediction explanations
backend/explainability.py
Model and evaluation
- ML ensemble
backend/part2_ml_model.py - Expanded training data
dataset/expanded_dataset.csv - Benchmark evaluation
backend/benchmark_eval.py
Other components
- Internship seeker
Connections 19
- Internship seeker → Analyzer interface: submits posting
README.md - Analyzer interface → FastAPI endpoints: requests analysis
- FastAPI endpoints → ML ensemble: predicts text
backend/main.py - FastAPI endpoints → Feature extraction: extracts features
backend/main.py - FastAPI endpoints → URL risk analysis: analyzes URL
backend/main.py - FastAPI endpoints → Payment detection: detects payment
backend/main.py - FastAPI endpoints → Prediction explanations: builds explanation
backend/main.py - FastAPI endpoints → Domain trust: checks trust
backend/main.py - ML ensemble → Feature extraction: extracts features
backend/part2_ml_model.py - ML ensemble → Expanded training data: trains on
backend/part2_ml_model.py - Feature extraction → Payment detection: checks payment context
backend/part1_dataset_similarity.py - Feature extraction → Text normalization: normalizes text
backend/part1_dataset_similarity.py - Payment detection → Text normalization: normalizes text
backend/payment_detector.py - URL risk analysis → Domain trust: uses brand domains
backend/url_checker.py - Prediction explanations → Feature extraction: explains features
backend/explainability.py - Benchmark evaluation → ML ensemble: evaluates predictions
backend/benchmark_eval.py - Benchmark evaluation → FastAPI endpoints: uses rule scoring
backend/benchmark_eval.py - Benchmark evaluation → URL risk analysis: scores URLs
backend/benchmark_eval.py - Benchmark evaluation → Feature extraction: checks features
backend/benchmark_eval.py