ML Engineer with expertise in GenAI, Multimodal AI, and Computer Vision. I built LensToWords — an image captioning system trained from scratch with 7 architecture iterations, achieving BLEU-4 of 0.1341 on COCO dataset. Currently researching how web agencies handle accessibility compliance at scale, specifically around automated alt text generation for client websites. Writing about web accessibility, ADA compliance, and the intersection of computer vision and real-world developer workflows. Connect with me: LinkedIn: linkedin.com/in/huzefa-merchant GitHub: github.com/huzefa10