Personal projects · Paul Kalaba

Casseo

A forever-free offline audio player. Smart categories sort your music by how you listen, and the whole player is yours to customize.

Free on Google Play

Casseo home screen: search, a Shuffle button for 1931 tracks, History and Favorites, and the most played albums.

Casseo Lists screen: playlists as round dot-art covers, with a red button to create a new list.

Never listened, a smart category of 30 tracks with Play and Shuffle buttons.

Casseo now playing screen with album art, a progress bar, and playback controls.

//dynodocy

Dynamic documents were a pain in the back, and no SaaS could fix that for me. With //dynodocy, you can upload a DOCX directly and get clean HTML in a second, ready to print to PDF. Pagination, tables, headers, and every other feature carry over automatically. Add your variables, and your dynamic template is ready.

The exported HTML file, with page size, margins, and print rules written as plain CSS.

The //dynodocy editor: a formatting toolbar above a paginated patent description template.

Page setup in //dynodocy: header and footer, page numbers, document fonts, and how images are exported.

DOCH app

Frankfurt is a small town, and it can get a bit boring. I lead a community of 600+ foreigners here, so I built DOCH: a calendar of every event worth going to and a guide to the places we recommend, all in one app.

The DOCH calendar for September 2025, with event days highlighted and the 29th selected.

A DOCH place card for KIKI LUCKY, a matcha spot in Frankfurt, with its address and an Open in Maps button.

Portfolio optimizer

An open-source portfolio optimizer. Pick a method (Markowitz, Max Sharpe, or Black-Litterman), then rebalance and backtest the result.

The Assets step: a filtered universe of stocks and ETFs with prices and dividend yields, beside sliders for capital, risk tolerance, portfolio size, and weight limits.

The Optimize step: Sharpe ratio, pure return, minimum volatility, and Black-Litterman side by side, each with expected return, volatility, and weights.

The Backtest step: $50,000 grows to $148,920 from 2014 to 2024, an 11.2% CAGR, with holdings and rebalancing costs.

Kalabot

Before I delegate anything to an AI version of me, it has to do the job right and stay out of the uncanny valley. Kalabot runs on Qwen with a behavioral layer that reads each situation (who’s asking, the hour, what’s already waiting) and replies the way I would.