Muhammad Zikrullah Rehman

PROFESSIONAL PROFILE Computer Science undergraduate at NUST (CGPA 3.64/4.00) with strong hands-on experience in deep learning, explainable AI, and biomedical signal analysis. Proven researcher with work spanning EEG interpretability, remote physiological measurement (rPPG), and physics-informed neural networks for real-time 3D brain source localization. Former AI Engineer at CVision, contributing production-level agentic AI systems and technical research briefings. Submitted research in XAI and document analysis, with international research exposure at Zhejiang University. Technically versatile across modern ML stacks, signal processing, and full- stack AI systems, with a strong focus on building interpretable, high-impact intelligent systems. EDUCATION Bachelors in Science, Computer Science SEECS , Islamabad , 3.64/4.00 (4) INTERNSHIP EXPERIENCE CVision 01-Apr-2025 - 31-Oct-2025 AI Engineer Islamabad, Pakistan – Promoted from research/analyst intern to part-time AI Engineer based on strong performance and technical contributions. – Designed and implemented AI agentic backend components, improving feature performance in company products. – Conducted comparative analyses of multiple backend features in clients’ web apps, recommended the optimal solutions. – Authored clear, concise technical briefing documents complete with flow diagrams, written explanations, and references to accelerate the dev team’s ramp-up on new features. – Researched and distilled emerging AI trends into briefs, producing LinkedIn infographics and posts that boosted company page engagement by 247%. Zhejiang University 30-Jul-2025 - 29-Aug-2025 Research Intern Hangzhou, China – Investigated deep learning approaches for remote photoplethysmography (rPPG) from video, focusing on robustness under varying illumination and motion conditions. – Implemented and benchmarked State-of-the-art spatiotemporal neural architectures designed for physiological signal extraction on different versions of rPPG datasets. – Conducted extensive experimentation, analyzing performance trends and extracting insights from evaluation metrics to identify strengths, weaknesses, and patterns in different model families. TUKL-NUST R&D Center 03-Jun-2024 - 28-Feb-2025 Research Intern Islamabad, Pakistan XAI Interpretability for EEG Classification: – Conducting research on explainable AI (XAI) algorithms such as SmoothGrad, Deep Taylor Decomposition, and Occlusion Sensitivity to make EEG classification models interpretable. – Evaluating the performance of these algorithms relative to a base EEG classification model. – Investigating the computational efficiency and localization capabilities of these XAI techniques. – Utilizing localization information from XAI algorithms to create an ensemble of Support Vector Machine (SVM) regression models, each trained on data generated by individual XAI methods, enhanced predictive performance by 34.1%. – Pioneered a new XAI method for EEG classification using wavelet features which improved upon previous SOTA methods – Findings have been compiled into a research paper and submitted to ISBI 2026. Urdu Ligature Recognition: – Developed a deep learning-based system using a CNN and a GRU to recognize Urdu language ligatures captured through a digitizer tablet. Recognized the ligatures with an accuracy of 97.6% – Evaluated the impact of handwriting features like pressure, tilt, and stroke direction on the recognition performance. – The demonstration was presented at the 16th International Workshop on Document Analysis Systems (DAS 2024), held in Athens, Greece, and was highly appreciated by the research community. Gaze based annotation of EEG files for training Deep Learning based classifier: – Replicated the results of a

Computer Science undergraduate at NUST (CGPA 3.64/4.00) with strong hands-on experience in deep learning, explainable AI, and biomedical signal analysis. Proven researcher with work spanning EEG interpretability, remote physiological measurement (rPPG), and physics-informed neural networks for real-time 3D brain source localization. Former AI Engineer at CVision, contributing production-level agentic AI systems and technical research briefings. Submitted research in XAI and document analysis, with international research exposure at Zhejiang University. Technically versatile across modern ML stacks, signal processing, and full- stack AI systems, with a strong focus on building interpretable, high-impact intelligent systems.