Hi, I'm Vesa, a solution architect from Finland.
I try to make complicated software easier to live with. I’m most interested in software that helps people work better and makes data easier to understand. A lot of that comes down to reducing unnecessary complexity. Lately I’ve been focused on software transformation, AI/LLM applications, and performance.
Converts legacy Cobol programs into modern Java. A grammar-driven parser and multi-stage deterministic AST pipeline produce runtime-correct Java. LLM stages then modernize and optimize the output.
An architecture auditing tool that speeds up getting oriented in an unknown codebase. It combines static analysis, dependency scanning, code metrics, and structure extraction into an interactive terminal-based findings browser.
Web/mobile application for quickly viewing departure times for nearby public transport stops. Data fetched realtime from the HSL GraphQL API.
Application which charts the major macroeconomic indicators with an EU focus. Data is fetched every day from the BLS, FRED, and ECB APIs.
Neural networks with automatic differentiation and backpropagation in Java from scratch.
In game development WFC (Wave Function Collapse) is used for generating output based on a set of tiles and rules. This output can then be used for different purposes, like level or texture generation. Here I'm creating textures. This is a simple Java implementation of Wave Function Collapse algorithm where the rules are defined in code.
2D game with procedural level generation and dynamic area of sight calculation.
Raytracer in Java.
Multiprocessed raytracer in Python.
Animated wallpaper for Android with minimal battery usage.