Projects

Contents

Projects and research directions I lead or contribute to. The full list lives on GitHub @HiDomesticCat.

Security & Systems Research Link to heading

DCFS — Zero-Knowledge Encrypted Cloud File System Link to heading

A cloud file system on AWS serverless infrastructure, built around the idea that the server should never see your files. Files marked private are encrypted in the browser before they leave it: a master key (MEK) is derived from a separate encryption passphrase via PBKDF2, each file gets a one-time data key (DEK) that encrypts the contents with AES-GCM, and the DEK itself is wrapped under the MEK before being uploaded alongside the IV — so the server only ever holds ciphertext and a wrapped key.

Access is split across three roles (anonymous visitor, user, administrator), transfers use S3 pre-signed URLs, metadata lives in DynamoDB, and Cognito issues the JWTs used for authentication.

  • Stack: AWS Lambda (Node.js), API Gateway, S3, DynamoDB, Cognito, Web Crypto API
  • Source: aws_finalwork

DeepCASE Log Correlation & Anomaly Detection Link to heading

Semi-supervised correlation analysis over syslog event sequences. It learns relationships between events and flags anomalous patterns (for example SSH brute-force attempts), with the goal of reducing alert fatigue in security operations. Supports rsyslog, remote hosts and file sources, and can deliver alerts to the console, email, webhooks or Slack.

SpectraLens — Photo Provenance Verification Link to heading

A proof of concept for verifying that a photo really was taken by this device, at this location. A Flutter capture app on the front end talks to a FastAPI backend that issues signing challenges and validates uploads, chaining together WebAuthn/Passkey device binding, Google Play Integrity device attestation and geolocation data.

Still at PoC stage — the external verification calls are currently mocked.

WannaMock — File Encryption from Scratch Link to heading

An educational project implementing a file encryption pipeline in C: recursive directory traversal, binary file I/O, and a cipher that evolved from a simple Caesar shift to RSA with a hand-written fast modular exponentiation routine. The write-up documents the memory-overflow and file-corruption problems hit at each iteration.

For academic and teaching purposes only, released under GPL-3.0.

Algorithms, Machine Learning & Visualisation Link to heading

Grover’s Algorithm Visualiser Link to heading

An interactive teaching tool for quantum search. It walks through superposition, phase inversion and amplitude amplification step by step, letting you pick a target state and watch the probability concentrate on the search result. The backend simulates the actual quantum circuit with Qiskit, with an optional noise model.

  • Stack: React, TypeScript, FastAPI, Qiskit
  • Source: Grover-demo

Hypercube Routing Visualiser Link to heading

Compares five routing algorithms on hypercube networks (BFS, A*, beam search, Batcher sorting and more), replaying the search step by step and running statistical comparisons across dimensions. All third-party libraries are vendored locally so it runs offline.

  • Stack: React, custom design-token styling system
  • Source: Hypercube-demo

sushi-sync — Operating System Synchronisation Simulator Link to heading

Uses competition for seats in a sushi restaurant to demonstrate OS multi-threading, resource allocation and synchronisation primitives (mutexes, condition variables). Features a live floor plan, seat configuration, timeline scrubbing and OS-oriented metrics — throughput, turnaround time, wait time — exportable to JSON, CSV or plain text.

  • Stack: Svelte 5, Tailwind CSS, Rust, Tauri
  • Source: sushi-sync

PSO-ANN-XOR — Training a Neural Network with Particle Swarm Optimisation Link to heading

Solves XOR with a 2-2-1 neural network trained by particle swarm optimisation instead of backpropagation: 50 particles, up to 2000 iterations, early stopping once loss drops below 0.001, plus a convergence chart.

MNIST — Training and Serving a CNN Link to heading

Handwritten digit recognition with a convolutional neural network, focused on what happens after training: exporting to SavedModel, serving REST and gRPC through TensorFlow Serving, packaging CPU and GPU environments with Docker, and training on GPU while exporting for CPU-only inference hosts.

  • Stack: TensorFlow, TensorFlow Serving, Docker
  • Source: MNIST

Applications & Coursework Link to heading

Product Search System Link to heading

A database course project: a full-stack product search and shopping cart application supporting keyword search, category browsing and cart operations. All database access goes through parameterised queries to prevent SQL injection.

Sites & Notes Link to heading

hicat0x0 Blog Link to heading

This blog. Built with Hugo and the hugo-coder theme, deployed to GitHub Pages. It collects CTF write-ups, security research notes and study logs.

Open Notes Link to heading

Course notes, learning resources and technical documents organised into a browsable site, published openly for classmates and the community.