Sabik Aftahee

CS Major | AI Engineer@SysModeler.ai | Blockchain Olympiad International Merit Awardee | 2x Hackathon Winner | Research: Medical Imaging, Computer Vision ,Trustworthy AI, NLP

I'm an AI engineer at SysModeler and a research assistant at Deakin University's School of IT, where I work with Dr. Akbar Husnoo. I studied Computer Science and Engineering at CUET in Bangladesh.

My interests are trustworthy AI, medical imaging and neuroscience. One question connects them: when can we rely on an AI system whose mistakes carry real costs?

Today's models score well on benchmarks, yet they still fail in ways that matter. A language model can answer a legal question fluently and still be wrong. An image classifier can reach the right diagnosis without using the evidence a doctor would look at. A report generator can describe findings the X-ray doesn't show. Most of my work so far catches failures like these: I measure reliability against human experts, test whether a model's explanations reflect real evidence, and ground what models generate in their input.

I also put this into production. At SysModeler, engineers upload aerospace and automotive specifications and get back SysML v2 requirements models that compile; the extraction pipeline I built cut hallucinated requirements from about 86% to under 5%.

These methods help, but they work from the outside, patching a system after it is built. For my PhD I want to study learning itself, starting with how a brain works out which earlier activity deserves credit when feedback arrives late, and what that can teach us about building more reliable AI.

I'm looking for a PhD position in NeuroAI or healthcare AI. You can reach me at sabikaftahee.official@gmail.com.

My CV and transcripts are linked here.