Mohammadi Tabassum

I’m an AI/ML engineer with hands-on experience building and deploying production LLM systems. I designed and shipped a multimodal RAG application (GPT-4o Vision + cross-encoder reranking) on AWS ECS Fargate, including streaming responses, page-level source citations, and a document-driven quiz room.

I also built a DNN-based fraud detection system with 99% accuracy on 50k records and an adaptive testing engine using Item Response Theory that dynamically adjusts difficulty in real time. Previously, as an AI & ML Intern, I improved CNN-based driver drowsiness detection accuracy by 12%, and as a trainee full stack developer, I developed Java backend modules for an event management platform with testing and documentation.