Freelance data scientist · All case studies

Mail that should never reach the inbox.

A spam and phishing classifier for corporate mail: TF-IDF and bag-of-words preprocessing, then a fair fight between multinomial Naive Bayes, SVM, and a CNN — evaluated, not declared.

Client: Corporate mail filter. Built by Dilshad Raza.

Corporate mail mixes invoices, attachments, and social-engineering copy. A single algorithm story is marketing. The work needed shared preprocessing and a comparison: classical text models against a small CNN, with metrics that survive a security review.

I built the preprocessing track — tokenisation, bag-of-words, TF-IDF — then trained multinomial Naive Bayes, an SVM, and a CNN, and evaluated them side by side so the security team could see cost, recall on the bad class, and whether convolution earned its keep.

Hire Dilshad Raza for similar freelance data science, machine learning, and AI automation in the UK, United States, and Australia.