KAMIL MARUSZAK

>> SYSTEM.RESEARCHER.EXE | STATUS: [ONLINE] | LOCATION: Kingston Upon Hull, UK

Computer Graphics & 3D Vision Researcher | Research Intern at USTC | Co-Founder & CDO/strong>


I'm an upcoming research intern at USTC under Dr. Juyong Zhang. I recently graduated from the University of Hull with a first-class degree in Computer Science and have a strong foundation in machine learning and 3D computer vision.

As Co-founder of MyndMap, an ADHD-focused productivity app, I currently lead research exploring computer vision and machine learning approaches to improve the daily lives of people with ADHD.

SYSTEM.BIO

My undergraduate dissertation, supervised by Dr. Qingde Li, focused on Kolmogorov-Arnold Networks (KANs) for 3D object representation, emphasizing efficiency and interpretability in point cloud classification.

HOBBIES.EXE

Basketball: Passionate about playing and following professional leagues. It teaches teamwork and strategic thinking.

Game Development: Creating interactive experiences and experimenting with game mechanics, computer graphics, and UX design.

RESEARCH

PRIMARY INTERESTS

  • 3D Computer Vision
  • Geometric Deep Learning
  • Computer Graphics
  • Neural Representations
  • Point Cloud Processing

FEATURED PROJECT

Kolmogorov-Arnold Networks for 3D Object Representation

Honours Stage Project | Summer 2024 - Present

  • Exploring KANs to model complex 3D topological features for geometric data representation.
  • Integrated KANs with 3D datasets using Python and PyTorch.
  • Designed framework for improved feature extraction and benchmarked against traditional MLPs.

PROJECTS.EXE

A collection of open-source tools, personal libraries, and experimental computer graphics toys I build in my spare time.

OPEN SOURCE TOOLS & LIBRARIES

[ PYTHON / PYTORCH ]

KAN-3D-Utils

A lightweight utility library extending Kolmogorov-Arnold Networks for 3D point cloud processing and visualization. Built to simplify geometric deep learning workflows.


VIEW ON GITHUB >>
[ C++ / OPENGL ]

FastPoint Renderer

An experimental, highly optimized C++ header-only library for rendering millions of points in real-time with custom shader support.


VIEW ON GITHUB >>

INTERACTIVE CG TOYS

Real-time Signed Distance Field (SDF) rendered using a WebGL fragment shader via sphere tracing. Move your mouse over the canvas to shift the camera angles, and use the terminal sliders to adjust the math parameters.

Experiment 01: Procedural Implicit Surface Raymarcher

RESEARCH BLOG

[2025-01-15] • RESEARCH NOTES
Understanding Kolmogorov-Arnold Networks

Kolmogorov-Arnold Networks (KANs) represent a fascinating departure from traditional neural networks. By placing activation functions on edges rather than nodes, they offer unique advantages in interpretability and scaling...

READ MORE >>

EXPERIENCE.DAT

Work History

Co-Founder & Research Lead | MyndMap

May 2025 - Present

  • Spearheading AI research in computer vision and machine learning for an ADHD-focused app.
  • Defining technical roadmap for AI-driven features and behavior pattern recognition.

Computer Vision Research Intern | MyndMap

Feb 2025 - Apr 2025

  • Conducted systematic literature reviews and developed comparative analysis frameworks for CV approaches in ADHD interventions.
  • Investigated resource-efficient ML models for mobile deployment.

Machine Learning Demonstrator | University of Hull

Sep 2024 - Present

  • Delivered interactive ML workshops and provided one-on-one Python/ML guidance.

Education

B.Sc Computer Science (1st Class)
University of Hull | Graduated Jul 2025

Technical Skills

Python C++ C# PyTorch TensorFlow scikit-learn Git Flask NumPy

CONNECT.INI

Currently based in Kingston Upon Hull, UK.


Email: maruszakkamil1@gmail.com

Phone: 07849 500 805

LinkedIn: @kamil-maruszak

GitHub: donkamilo00

Portfolio: donkamilo00.github.io