I don't just research ML. I build it, ship it, and prove it works.

• PhD Computer Vision (UTM Malaysia, 2022), published CV architecture that improved mAP by 7% at zero additional compute cost
• 20+ publications, 571 citations, h-index 8
• Head of AI Department at IQRA University, running a team of 6 researchers, supervising 2 PhD + 3 Master's students
• Production systems: sub-100ms inference, RAG pipelines, LLM fine-tuning, medical imaging diagnostics

Published architecture that improves detection at zero extra cost

A novel feature pyramid refinement module that boosts object detection accuracy without increasing inference time. Published in Image and Vision Computing, 2021.

Explainable AI that clinicians actually trust

Attention maps, concept-based explanations, and audit-ready outputs for medical diagnostics. Breast cancer diagnosis, drug-response prediction, ECG/CVD signal analysis.

Production RAG that hallucinates less

Dense + sparse retrieval with citation grounding for verifiable LLM outputs. Built for production deployment.

Domain-adapt LLMs on a single GPU

Fine-tune 7B-70B parameter models with LoRA and QLoRA. Memory-efficient, production-ready pipeline.

Temporal analysis meets interpretability

Hybrid spatial-temporal architecture for deepfake detection with explainable outputs. FaceForensics++ benchmark.

Full list on Google Scholar →

ML Engineer / Research Scientist roles in Netherlands, Germany, or remote EU. Visa sponsorship required.

I thrive in teams that value shipping + research depth over headcount. I build what I publish, every repo ties to a paper with real numbers.