Experience

Machine Learning Intern

Industrial Practical Training

Emerging Technologies for Healthcare (ETH) Lab, Muhimbili University of Health and Allied Sciences (MUHAS)

Dar es Salaam, Tanzania

Building and deploying models against real clinical constraints — offline devices, explainability requirements, and datasets assembled across institutions.

// selected work

• Designed and deployed Mwana AI, a breast cancer ultrasound classifier (PyTorch, ONNX, Flutter) built for offline use in clinics.
• Built a CNN-based TB/HIV co-infection detection system with Grad-CAM explainability.
• Led multi-institution data extraction for a respiratory disease study across Aga Khan, Warwick, and NTLP.
• Assisted in fine-tuning the RETFound foundation model for diabetic retinopathy detection.
• Applied transfer learning to cardiac imaging for dilated cardiomyopathy prediction.
• Designed data preprocessing pipelines for medical imaging and audio datasets.

• PyTorch
• ONNX
• Flutter
• Grad-CAM
• Transfer learning
• Medical imaging