Muhammad Maaz Aleem

AI, ML & the future of medicine.

BSCS student and full-stack engineer moving into machine learning applied to healthcare — with current research focused on domain shift and out-of-distribution (OOD) generalization in clinical diagnostics, ensuring medical computer vision models do not fail silently when deployed across different hospital scanner vendors, acquisition protocols, and heterogeneous patient cohorts.

Current Research Direction: Investigating acquisition differences between Siemens, GE, and Philips imaging hardware, and developing Test-Time Adaptation (TTA) algorithms to safeguard diagnostic performance across real-world hospital networks.

Explore Research Focus Read Background & Roadmap View Full CV Get in Touch

Neural Architecture & Latent Inference

Interactive Forward Pass

Figure 1.0 — Architectural Flow: Domain-Invariant Clinical Latent ManifoldTest-Time Adaptation · Out-of-Distribution Calibration · Multi-Scanner Invariance

Forward Propagation (PyTorch JIT)|Loss: CE + Dice + IRM

Domain Alignmentz₃⁽¹⁾: Invariant Representation

Domain-adversarial feature alignment neutralizing scanner shift

Core Competencies & Focus Vectors

All 4 Pillars

Active Research

Domain Shift in Medicine

Investigating acquisition differences, cross-scanner covariate shifts, and test-time adaptation in clinical diagnostics.

Neural Modeling

Deep Learning Foundations

Autograd, backprop, custom loss functions, and vision architectures rebuilt from first principles in PyTorch.

Big Data Systems

High-Throughput Pipelines

FlyRank warehouse processing on tens of millions of records without memory exhaustion.

Degree & Location

Air University BSCS

Based in Islamabad, Pakistan. Class of 2028, balancing rigorous computer science with clinical AI research.

Explore Portfolio Sections

Independent Pages

SECTION 01