CV
Below is my updated CV, also attached in .pdf format.
Contact Information
| Name | Muhammad Turab |
| Professional Title | PhD Researcher |
| turabbajeer202@gmail.com | |
| Location | , Dublin, |
Professional Summary
I specialize in audio/speech recognition, image processing, LLMs, and agentic AI-based methods, with hands-on experience developing agentic AI applications for various use cases.
Experience
-
2025 - Present Dublin, Ireland
PhD Researcher in Explainable Medical Imaging
Dublin City University
Developing neuro-symbolic methods in medical imaging to make neural network decisions more explainable and transparent.
- Develop neurosymbolic-based methods for explainability and transparency in medical imaging.
-
2025 - Present Saint-Etienne, France
Master's Thesis Student
Laboratory Hubert Curien, University of Jean Monnet
Research on 3D pose estimation, 3D scene reconstruction, and movement analysis of dancers.
- 3D pose estimation and human body modeling of dancers.
- 3D scene reconstruction to estimate the floor on which dancers are moving.
- Movement evaluation using trajectories of joints/bodies.
- Developed novel feature descriptors for movement description and analysis.
- Evaluated feature descriptors using ML and explainability (SHAP, LIME).
- Investigated space usage, movement quality, and joint initiation.
-
2024 - 2024 Norway
Research Intern
Colourlab, NTNU
Colorization of old black-and-white negative films using generative AI (Pix2Pix-based GAN) and custom palette matching.
- Built a Pix2Pix-based GAN workflow for film colorization.
- Developed palette matching with super-pixel segmentation and color transfer.
- Reported performance surpassing DeOldify on tested material.
- Tested on a painting from the 1930s with promising results.
-
2022 - 2023 Ireland
Research Intern
University of Galway
Audio classification research: preprocessing, augmentation, multi-feature selection, and robustness improvements.
- Integrated 5 data augmentation methods; reduced overfitting by 15%.
- Developed multi-feature selection algorithm combining 5 spectral/temporal features; improved accuracy by 12%.
- Developed AudRandAug augmentation technique (weighted and non-weighted strategies).
-
2022 - 2022 Pakistan
Research Assistant
Department of Computer Systems Engineering, MUET
Research on forged character detection in personal documents and dataset generation.
- Worked on forged character detection for passports and driver’s licenses.
- Developed synthetic datasets and ML/DL methods for forgery detection.
- Released three public datasets (15,000 images each).
Education
-
2024 - 2024 Oxfored, UK
Summer School
University of Oxford and AI for Global Goals
Oxford Machine Learning Summer School
- Statistical and Probabilistic Machine Learning; Representation Learning (Generative AI); Geometric Deep Learning
- Gained insights into Computer Vision, NLP, LLMs, Knowledge Graphs, and Neuro-Symbolic AI.
-
2023 - 2025 Norway, France, Finland
Joint Master's Degree
Erasmus Mundus Joint Masters Degree Program
Computational Color and Spectral Imaging
- Image Processing; Machine Learning; Computer Vision; Deep Learning; Color Science; Spectral Imaging; Hyperspectral Imaging
- Relevant coursework in imaging, ML, and spectral/color science.
-
2018 - 2022 Hyderabad, Pakistan
B.E.
Mehran University of Engineering & Technology (MUET)
Computer Systems Engineering
- OOP; Databases; Data Structures & Algorithms; Operating Systems; DBMS; Web Engineering; Machine Learning; Data Mining; Advanced Data Structures & Algorithms; Image Processing
- Undergraduate degree in Computer Systems Engineering.
Awards
Publications
-
2025 Movements analysis in performing arts from human body pose estimation
Proceedings of the 2nd International Conference on Signal Processing and Computer Vision (SIPCOV-2025) — invited paper (to be published in Springer Nature)
-
2025 Emotion Recognition in Contemporary Dance Performances using Laban Movement Analysis
Submitted to: 21st International Conference on Computer Analysis of Images and Patterns (CAIP)
-
2025 Dance Style Recognition using Laban Movement Analysis and Machine Learning
Submitted to: Advanced Concepts for Intelligent Vision Systems (ACIVS)
-
2023 A Comprehensive Survey of Digital Twins in Healthcare in the Era of Metaverse
BioMedInformatics, 3(3), 563–584
-
2023 -
2022 Investigating Multi-Feature Selection and Ensembling for Audio Classification
International Journal of Artificial Intelligence & Applications
-
2022 FORGED CHARACTER DETECTION DATASETS: PASSPORTS, DRIVING LICENCES AND VISA STICKERS
International Journal of Artificial Intelligence & Applications
-
2023 An Ensemble of CNN Architectures for Early Detection of Alzheimer’s Disease Using Brain MRI
Mehran University Research Journal of Engineering & Technology
-
2022 Data Dimension Reduction Makes ML Algorithms Efficient
International Conference on Emerging Technologies in Electronics, Computing and Communication (ICETECC) 2022
Skills
Languages
Interests
Certificates
- Deep Learning Specialization - DeepLearning.AI on Coursera (2024)
- Machine Learning Specialization - University of Washington on Coursera (2023)
- Data Science Specialization - Johns Hopkins University on Coursera (2023)
Projects
-
3D Reconstruction of Static Scenes from Multiview Images
3D reconstruction pipeline using multiview geometry and deep learning baselines.
- Camera calibration, SIFT+FLANN matching, stereo rectification, disparity mapping, depth estimation.
- Used Semi-Global Block Matching (SGBM) for improved disparity maps.
- Experimented with classical methods and DUST3R.
- Generated 3D point clouds and compared multiview results.
-
Real-time Virtual Object Insertion via Depth Estimation and Environment Mapping
Realistic object insertion using depth estimation, environment/light estimation, and inpainting.
- Used DistDepth, DPT, DepthAnything, ZoeDepth.
- Depth + lighting for scaling/placement realism.
- Iterative inpainting and DiffusionLight for integration realism.
- Improved performance and visual quality vs traditional techniques.
-
Image Signal Processing Approaches for Low-Illumination Image Enhancement
Survey/review of low-light image enhancement methods.
- Analyzed traditional denoising and automatic white balancing.
- Reviewed deep learning approaches (CNNs, GANs) and hybrid methods.
- Compared strengths/limitations and suggested future directions.
-
Image Super-Resolution using Generative Adversarial Networks (GANs)
SRGAN-based super-resolution for perceptually realistic image enhancement.
- Implemented content + adversarial losses.
- Trained with simulated degradations and high-resolution datasets.
- Improved sharpness/detail vs interpolation baselines.
-
Spectral-Based Contrast Enhancement and Camera Simulation for Biological Imaging
Spectral contrast enhancement and camera simulation using biological spectral datasets.
- Worked with Dental Spectroscopy Database.
- Simulated camera spectral response and applied spectral estimation.
- Enhanced contrast for skin/mucous membrane imaging; endoscopy-focused evaluation.
-
Smart Contact Manager
Secure web application for contact management using Spring Boot.
- CRUD, signup/signin, email OTP password recovery.
- Spring MVC/Security, JPA, Thymeleaf + Bootstrap UI.
-
City Mall Management System
Desktop shopping mall management software.
- Account/product/order modules; purchase logging; transaction history.
- Role-based access control; Java Swing UI.