I’m a Computer Science student focused on AI and DevOPS.
I enjoy building things from scratch — neural networks, system-level tools, and full-stack applications — to deeply understand how they work under the hood.
- Machine Learning & Deep Learning
- Linux internals and tooling
- AI Systems & Optimization
- Cybersecurity and Penetration Testing
- Studied Deep Neural Networks for solving non-local conservation laws derived from the Kuramoto model.
- Analysed the impact of activation functions, depth, width, and training setup on convergence and accuracy.
- Compared PINNs with classical numerical methods, highlighting accuracy trade-offs and limitations near discontinuities.
- Investigating alternatives to backpropagation using Direct Feedback Alignment (DFA).
- Introduced ViT-DFA, applying DFA-based learning to Vision Transformers.
- Evaluating convergence, robustness, and generalisation on medical image classification tasks.
- Ricing my linux for programming efficiency
- Working on research-oriented ML projects
- Improving system design and engineering fundamentals
- GitHub: https://github.com/m4YnK-7
- LinkedIn: https://linkedin.com/in/mayank-singh-patwal