I'm a curious person with broad interests, always eager to dive into intellectually challenging projects that push my learning boundaries. I love working on ideas that have the potential to create a real-world impact.
I approach each project with a deep curiosity about how things work and how they can be improved. What excites me most is the potential of curiosity to spark innovation and push the boundaries of understanding. I typically start a project when I feel a βscratching itchβ about the current way a system is laid outβwhen something feels off or inefficient. That itch drives me to reimagine the problem, exploring new perspectives and creating more effective solutions.
One of my passions is exploring and studying the design of programming languages. Iβm particularly fascinated by how language design influences the way we think about problems and develop solutions. Some of my favorite languages from a design perspective include Haskell, LISP, REBOL, Pony, E, Erlang, Python, Smalltalk, and Zig.
I also enjoy tinkering with distributed systems and novel data structures (have you seen the type of prolly trees used by Dolt?), as well as exploring cutting-edge deep learning research, which is the focus of my PhD.
In terms of software, I have a deep appreciation for stable and well-crafted tools. Some of my favorites include SQLite, Emacs, Kubernetes, Fossil SCM, and pretty much everything created by Fabrice Bellard. I also follow with great interest the experimental projects coming out of Ink & Switch. Beyond these, my favourite frameworks are SvelteKit, Django, and PyTorch for their unique take on their intended usage.
I like to contribute to open source projects by providing feedback, suggestions, and solutions through issue discussions. One of my significant contributions has been in the ZeroNet project, which revolves around decentralized websites.
This section includes research-related projects that are part of my PhD in the study of Deep Learning Optimization Algorithms.
- BMVC2022 β A simple website that collects code, slides, poster, and the paper presented at BMVC2022.
- ADLER (Adaptive Learning Rate with Approximated Hessian) β Derives a positive semi-definite Hessian approximation for deep learning models to derive an optimal learning rate, compared with a grid search on SGD.
- Multi-PL Networks β This repository hosts the experiments performed to test whether real Deep Networks behave like their abstract modeling via Polyak-Lojasiewicz conditions having a PL constant per layer.
These are a sample of the various small projects that I have produced over the years which can be considered in a stable working state:
- StylusLabs Write Export β Python script to export StylusLabs Write SVGZ files to PDFs with functional links.
- Obsidian Hotkeys Chords β An Obsidian plugin for setting hotkey chords (key sequences) to trigger commands, allowing for Emacs-like keybinding flexibility.
- SQLite Under Git β A setup for handling SQLite databases in Git, transforming them into textual form using git clean/smudge filters during commits.
- Template Docker CI Image Builder β Demonstrates how to build a Docker image using GitLab CI and upload it to the GitLab Container Registry.
- Multicast Chat β A terminal-based chat application that uses multicast messaging, written in Haskell.
- Buildpack LaTeX β A custom Heroku-style buildpack setup for building LaTeX documents using specified document classes.
- Scherzi β A collection of computer pranks, perfect for when someone leaves their account unlocked.
- PWGen Listing Tool β Modified pwgen that generates all possible passwords using a backtrack programming utility.
Throughout my studies, I have assembled several materials and notes on various mathematical subjects, which you can find below.
- Mathematical Notes and Links β A collection of notes and resources that I have assembled during my studies in mathematics.
- FattiDiMaterie β Various facts and handouts (styled as summaries) on different subjects. Some are incomplete or still in draft form.
- ProblemiInteressanti β A collection of interesting mathematical problems. This repository mainly contains my explorations and thoughts on different topics.
These projects, while not entirely complete, have been essential in my journey to learn and experiment with different technologies and stacks:
- Binary Diff β A diff-like and patch program designed for binary files to compare and analyze data.
- CRC Bruteforce β Tool to brute-force CRC generating polynomials using known strings and their CRC values.
- Node.js Multicast Experiment β Experiments with UDP multicast using Node.js for a chat application.
- Org2Book Converter β Converts modified org-mode files into LaTeX for structured document creation.
- GEPy (Gene Expression Programming) β Implementation of Gene Expression Programming in Python3, still in alpha stage.
- Maxtract Docker β Docker image for converting PDF files to LaTeX source code with Maxtract.
- UOWHR (Uguale od Opposto Whole History Rating) β Skill ranking system based on the Whole History Rating algorithm for sport players.
- AsteriskCentralina β Software for creating a customizable telephone exchange system using Asterisk.
- Telegram Bot Experiment β A simple Telegram bot designed to test the features of the Telegram Bot API.
- Cloudflare Example Application β SvelteKit app example for deploying to Cloudflare Pages.
- Basic Simple Jekyll Site β A simple, pre-configured Jekyll site with a range of built-in facilities.
- Isolah Game Manager β Code for the Game Manager of the real-time game "Isolah", built with Django.
- Blackbird WebOS β A non-framework web approach that uses a thin JavaScript client communicating with a backend via WebSocket for clean and efficient code.
- TBT - The Bash Templater β A straightforward bash script for templating files with ease.