Software and Web Development Student | Systems, IoT & Deep Learning

I am studying Web Development at IFRS. While my formal coursework focuses on full-stack web applications, I am a self-learner driven by low-level systems, hardware, and AI.

• Cybersecurity & Infrastructure: Focused on networking, cryptography, and DevOps. Developed a custom cryptographic algorithm and actively manage a home server using Docker and Tailscale for secure remote access.
• IoT & Embedded Systems: Worked for two years at the FabLab-affiliated laboratory on campus, gaining hands-on experience with microcontrollers using ESP-IDF and Arduino frameworks.
• Deep Learning: Certified by the NVIDIA Deep Learning Institute (DLI). Built neural networks from scratch, trained models, and tuned hyperparameters, backed by a solid understanding of the underlying mathematics.

A lightweight 2D rendering and linear algebra engine for the browser, built on top of the HTML Canvas API using TypeScript.

• Tech Stack: TypeScript, HTML5 Canvas.
• Key Architecture:
  • Custom Math Engine: Built-in immutable 2D vectors (Vector2) and 3x3 transformation matrices (Matrix3) for handling affine transformations.
  • Hierarchical Scene Graph: A Transform system supporting parent-child object hierarchies, computing world matrices down the chain.
  • Polled Input Management: A frame-based InputManager that abstracts mouse, touch, and keyboard interactions for real-time applications.
  • High-Level Renderer: Encapsulates the native canvas context, offering a strongly-typed API for geometry, text, and state management.

A custom pure JavaScript cryptographic library that integrates multiple security techniques (obfuscation, bitwise, CBC, non-linear PRNG) to protect messages. Designed as an educational case study and not for production use.

• Tech Stack: Pure JavaScript (ES6+), BigInt, Web Crypto API (crypto.getRandomValues).
• Key Architecture:
  • Custom PRNG (Xorshift128 Modified): Developed a modified, non-linear Xorshift128 pseudo-random number generator that integrates state addition, dynamic bitwise rotation, and modular arithmetic to resist linear algebra attacks.
  • Key Derivation & Stretching: Employs a derived 128-bit hash (gerarHash) using 100,000 iterations of FNV-1a with BigInt, combined with a randomly generated 48-byte Initialization Vector (IV) to prevent rainbow table collisions.
  • Encryption Pipeline: Implements a multi-layered process, starting with a block cipher in Cipher Block Chaining (CBC) mode, followed by bitwise obfuscation through dynamic garbage bit injection, resulting in a highly non-deterministic ciphertext.
  • Authentication (Encrypt-then-MAC): Generates a 32-character hexadecimal MAC from the final ciphertext for message authentication. Critically, it includes a constant-time comparison function (constantTimeCompare) to mitigate timing attacks during decryption.

A lightweight 3D software wireframe renderer built from first principles using vanilla JavaScript and the Canvas 2D API.

• Tech Stack: Vanilla JavaScript (ES6+), HTML5 Canvas.
• Core Architecture:
  • Custom Linear Algebra Library: Implements raw Matrix storage and transformation algorithms (MatrixMath) from scratch, handling standard matrix multiplication, transposition, and Hadamard products.
  • Graphics Pipeline Mathematics: Manually processes 3D vertex transformations using X and Y axis rotation matrices alongside 3D coordinate translations.
  • Perspective Projection: Simulates depth and Field of View (FoV) through analytical division mechanics (fov / (fov + z)), projecting 3D environments onto a 2D screen viewport.
  • Centroid-Based Scaling: Dynamically computes geometric centroids for arbitrary polyhedral meshes to isolate uniform scaling relative to the object's center of mass.

• Email: caetanogoncalves@proton.me