An open-source AI workspace alongside university projects in C, Python and RISC-V. Together they cover reusable workflows, data structures, memory management and translating specifications into working software.
An open-source Obsidian workspace for organizing knowledge, running projects and reusing skills across AI agents, with one shared set of instructions.
I designed the system and guide its development with AI assistance: portable skills, guided setup and Python tools for workspace health checks and release validation.
- Obsidian
- Python
- AI workflows
- open source
View Project A command-line invoicing system in C: products, stock, carts, VAT classes, clients and invoices, all held in memory and manipulated through a parsed command language.
Products live in a hash table keyed by EAN with linked-list collision handling; clients in a binary search tree. Includes EAN-8/EAN-13 and NIF validation, manual memory management and Doxygen docs. Passes all 31 automated tests.
- C
- terminal app
- hash tables
- BST
- validation
View Repo
Mini-Transformer in RISC-V
A next-token prediction pipeline written in RISC-V assembly. The attention mechanism reduced to registers, memory addressing and loops.
Vocabulary lookup, embedding tables, Q/K/V matrix operations and attention scoring for the last token, with no abstractions to hide behind. Graded 20/20.
- RISC-V
- assembly
- Q/K/V
- matrices
- memory
View Repo A Scrabble-style word game in Python for 2-4 local players, with a 15x15 board, tile bag, scoring and bot opponents that play from a vocabulary file.
Turning a long ruleset into testable functions: board state, placement validation, scoring, seeded shuffling for reproducible games, and a public pytest suite.
- Python
- terminal game
- bots
- validation
- game logic
View Repo 
