Humera Tariq

AI/ML researcher, computer scientist, and educator with 20+ years of experience across medical imaging, computer vision, deep learning, data science, scientific software, and academic leadership. International postdoctoral research at the University of Michigan and the University of Sassari; strong record of building reproducible computational pipelines, developing and evaluating 2D/3D models, collaborating across clinical and technical teams, and communicating complex work through publications, teaching, and invited talks.

Assistant Professor, Computer Science at University of Karachi (2024-01 – Present)

Department of Computer Science

• Teach and design advanced courses in machine learning, deep learning, image processing, computer graphics, software engineering, human-computer interaction, and web development, using continuous assessment and industry-oriented projects.
• Supervise student research and contribute to institutional quality and accreditation activities aligned with NCEAC and HEC requirements.
• Conduct prompt-engineering and responsible-AI training for university faculty as a Meta AI Master Trainer in collaboration with Atomcamp, NCEAC, and HEC.
• Maintain research collaborations in medical imaging, including AI-assisted infarct-core and penumbra detection/segmentation from NCCT and DWI.

Postdoctoral Research Fellow, MRI & Deep Learning at University of Michigan (2022-01 – 2024-12)

Department of Radiology

• Configured and administered an NVIDIA RTX A6000 GPU research environment and built a reproducible PyTorch pipeline for quantitative MRI segmentation.
• Engineered secure data flow from institutional imaging servers, organized inputs and outputs at patient/series level, and produced standardized segmentation results for downstream analysis.
• Developed and compared 2D U-Net, 2D Attention U-Net, 3D U-Net, and 3D Attention U-Net models for proximal femur and posterior ilium bone-marrow segmentation in myelofibrosis MRI.
• Implemented training, validation, held-out testing, hyperparameter optimization, post-processing, visual QC, and quantitative evaluation using Jaccard, volume-intersection, volume-error, Hausdorff-distance, and related metrics.
• First author of a Frontiers in Oncology paper and contributor to murine tibia segmentation validation, test-retest analysis, and a University of Michigan Deep Blue data/software release.

Postdoctoral Research Fellow, Biometrics & Recognition at University of Sassari (2020-01 – 2022-12)

Computer Vision Laboratory

• Researched biologically inspired face and periocular recognition under the SPADA secure digital-identity project.
• Designed and evaluated log-polar representations, HMAX features, PCA/ICA, CNNs, autoencoders, feature fusion, contrastive-learning concepts, and cosine-similarity-based verification in Python and MATLAB.
• Processed and analyzed Caltech, OULU, FRGC, and UPFR biometric datasets; prepared technical reports, presentations, and statistical identification/verification analyses.
• Delivered online lectures and hands-on instruction on GPU architectures and CUDA at short notice during the COVID-19 remote-work period.

Permanent Faculty Member, Computer Science at University of Karachi (2006-01 – 2020-12)

Department of Computer Science

• Taught undergraduate and graduate courses across programming, algorithms, image processing, computer graphics, software engineering, HCI, machine learning, and systems development.
• Supervised research and capstone projects in medical-image analysis, brain MRI skull stripping, emergency-department process simulation, computer graphics, and software automation.
• Served in student advising, examination-record management, ORIC/IAESTE coordination, curriculum and quality activities, and industry-academia collaboration.

Visiting / Cooperative Teaching Faculty at University of Karachi and partner institutions (2000-01 – 2006-12)

MAJU, KIET, and technical institutions

• Delivered computer-science and software-development teaching across university and professional-training settings.

AI-assisted stroke imaging at Dow University of Health Sciences (2026-01 – Present)

Developing methods to assist detection and segmentation of infarct core and penumbra from NCCT and DWI, with emphasis on reliable analysis, uncertainty-aware evaluation, and clinically meaningful outputs.

Oncology clinical-trial replication at Keiji.ai TrialMind Research Camp (2026-01 – 2026-12)

Participated as a data scientist in real-world evidence replication of DESTINY-Breast03 and CLEOPATRA, plus GBM bioinformatics work; produced cohort analyses, survival results, visualizations, target-discovery outputs, and QC.

Ph.D. in Computer Science – University of Karachi, Pakistan (2012-01 – 2015-12)

MS/M.Phil. in Computer Science – University of Karachi, Pakistan (2009-01 – 2012-12)

Master of Computer Science in Computer Science – University of Karachi, Pakistan (2001-01 – 2003-12)

B.E. in Electrical Engineering – NED University of Engineering and Technology, Pakistan (1995-01 – 1999-12)