I'm developing the RBA AI Email Classifier and Assistant, which routes incoming emails by intent and category and matches them to relevant FAQs.
I built confidence scoring and human-review recommendations into the project, so uncertain matches can be flagged rather than answered automatically. During development, the automated regression suite reached 39/39 tests passed.
Through iterative testing, I identify classification failures and refine the system, including distinctions between closely related intents such as tax on contributions and tax on retirement payouts.
I've also worked with Google AI Labs and generative AI tools on comic-book content, refining generated concepts and outputs for quality, relevance, and consistency.
