Dr. Amir Salarpour

PostDoctoral Researcher at Clemson University | Assistant Professor & Thesis Supervisor | Focus: Computer Vision & Deep Learning

Dr. Amir Salarpour
Current Affiliation:
PostDoctoral Researcher
Clemson University
(Former Assistant Professor at Sirjan Univ. of Tech)

Biography & Collaboration

Dr. Amir Salarpour is a distinguished researcher specializing in Computer Vision and Deep Learning. I had the profound privilege of working closely under his direct supervision for two transformative years (February 2019 to February 2021) at Sirjan University of Technology.

During this extensive collaboration, I served in multiple capacities simultaneously: as his Undergraduate Thesis Student, a dedicated Research Assistant (RA) in his laboratory, and a Teaching Assistant (TA) for foundational and advanced programming courses.

Under Dr. Salarpour's mentorship, I cultivated a deep-rooted philosophy of understanding machine learning architectures from their mathematical foundations rather than relying solely on high-level APIs. His guidance was instrumental in transitioning my skills from classical algorithms to state-of-the-art Deep Learning models handling unstructured 3D data.

"She became an integral member of my team... consistently exceeding my highest expectations. Unlike many students who rely solely on pre-built libraries, Ms. Amini insisted on mastering the underlying mathematics. She is, without a doubt, one of the most capable undergraduate students I have mentored."

Key Research & Technical Achievements

3D Point Cloud Processing & Research (RA)
  • Served as a Research Assistant focusing on complex, unstructured, and unordered 3D data requiring permutation-invariant architectures.
  • Developed and implemented novel ideas based on Kernel Correlation for Local Feature extraction.
  • Successfully reproduced and benchmarked state-of-the-art architectures including PointNet and DGCNN on the standard ModelNet40 dataset using Python, PyTorch, and TensorFlow.
Thesis & Advanced Coursework
  • Thesis Excellence: Authored the undergraduate thesis titled "Large-scale Semantic Point Cloud Segmentation with Superpoint Graphs," earning a perfect score of 20/20.
  • Algorithmic Mastery: In the "Speech and Language Processing" course, implemented complex algorithms like Logistic Regression and Neural Networks (including full Backpropagation) entirely from scratch to master the underlying mathematics.
  • Signal Processing: Developed a "Spoken Digit Recognition" system utilizing MFCC feature extraction and Gaussian Mixture Models (GMM) for high-accuracy audio classification.
Teaching Assistantship & Leadership (TA)
  • Served as a TA for "Basics of Programming" and "Advanced Programming" for two years, teaching complex C/C++ concepts including Memory Management (Pointers) and OOP principles (Polymorphism, Inheritance).
  • Played a pivotal leadership role during the 2020 COVID-19 pandemic by managing online classes, conducting precise Code Reviews, and enforcing Clean Code practices.

Official Recommendation Letter

Document Verification

Status: Officially Signed
Issued By: Dr. Amir Salarpour
Institution: Sirjan University of Technology
Endorsement Highlights: Exceptional mastery of underlying mathematics (implementing Backpropagation from scratch), pioneering research in unstructured 3D data (PointNet, DGCNN), and critical leadership during the shift to remote education.