Mr. Dildar Shah SCIT - BNU

Mr. Dildar Shah

Lecturer

School of Computer & IT (SCIT)

Dildar Shah is a Lecturer at SCIT, Beaconhouse National University, Lahore, specializing in optimization-based machine learning, model compression, and efficient deep learning for edge systems. With a Master’s in Applied Mathematics from GIKI, his research focuses on pruning, compression-aware training, and resource-efficient AI using mathematical optimization and probabilistic modeling.

Bio

Dildar Shah is a Lecturer at the School of Computer and Information Technology (SCIT), Beaconhouse National University (BNU), Lahore. He is a mathematician and applied researcher specializing in optimization-based machine learning, model compression, and efficient deep learning for embedded and edge systems.

He holds a Master’s degree in Applied Mathematics from Ghulam Ishaq Khan Institute (GIKI) and has a strong background in mathematical optimization, signal processing, and probabilistic modeling. His research focuses on train-time network pruning, compression-aware training, and developing resource-efficient deep learning models.

He is proficient in Python-based ML frameworks such as PyTorch and TensorFlow and actively works at the intersection of mathematics and scalable AI systems.

Academics

  • MS Applied Mathematics, Ghulam Ishaq Khan Institute (GIKI), Pakistan
  • International Mathematics Master (IMM), ICTP Affiliate Program, COMSATS Lahore
  • BS Mathematics, University of Peshawar, Pakistan

Experience

  • Lecturer, Qurtuba College, Hayatabad (2021–2022)
  • Graduate Assistant, GIKI (Linear Algebra, Calculus, Probability & Statistics)
  • Student Representative, Maths Volunteers Foundation (2023–2024)
  • Member, Interdisciplinary Machine Learning Initiative (IMLI), GIKI

Research Projects

  • Efficient Neural Network Pruning for Edge Devices – ResNet, MobileNet, WideResNet models with structured and unstructured pruning achieving up to 87% sparsity and reduced training time
  • Quaternion-Based Sparse Signal Recovery (Ongoing) – Collaboration with KFUPM on compressive sensing using quaternion models and Bayesian methods

Publications

  • Shah DI, Hanif M, Butt N. (2025). Variance-Guided Structured Pruning for Optimized CNNs. ICCSA 2025 (Springer)
  • Unstructured Adaptive Feature Pruning for Optimised CNNs using Taylor Expansion (Under Review, IEEE TETCI)
  • Additional publications in Chaos, Solitons & Fractals (2021) and Differential Equations and Applications (2023)

Technical Skills

  • Python, NumPy, Pandas
  • PyTorch, TensorFlow, Scikit-learn
  • Deep Learning: CNNs, RNNs, Transformers
  • Signal Processing: PCA, SVD, Wavelets
  • Optimization: Convex Optimization, Stochastic Methods, PSO, GA
  • Machine Learning Theory & Bayesian Inference

Presentations

  • ICCSA 2025 Presentation – Variance-Guided Structured Pruning for Optimized CNNs
  • MS Thesis Defense – Efficient Neural Network Pruning for Embedded Systems (GIKI)
  • Introduction to Probability and Stochastic Processes – Maths Volunteers Foundation (2023)

Awards & Certifications

  • MS Scholarship – GIKI (2023)
  • IMM Scholarship – COMSATS/ICTP (2022)
  • Merit Certificate – BS Mathematics (2020)
  • Coursera: Crash Course on Python – Google (2024)

Contact

© 2026 BNU.
Powered bytossdown.com