AI/ML Engineer Intern Ezitech Institute
Fine-tuned a BERT-based fake news detection model using Hugging Face Transformers, achieving approximately 93% accuracy on more than 5,000 labeled news articles.
Built a phishing website detection classifier using a Scikit-learn pipeline for feature extraction and evaluation, reaching approximately 91% accuracy, a 0.90 F1-score, and 0.95 ROC-AUC.
Constructed an end-to-end Comic Story Generator combining Groq LLM narrative generation with Coqui TTS voice narration, reducing manual content creation time by approximately 70%.