Welcome to my minimalist blog featuring my thoughts, projects, and opinions on technical topics.
Announcing Auto-Grade v0.2.0: a milestone release featuring complete file management, polished UI, Python 3.13 upgrade, and the foundation for AI-powered grading. A reflection on the journey from hackathon prototype to production-ready platform.
A deep dive into building a LibreUtils tool from scratch — separating pure logic from UI, establishing patterns that scale, and creating a text encoder that handles Base64, URL encoding, and six other formats.
How we integrated OCR, LLMs, and vector embeddings to transform Auto-Grade from a file management system into an intelligent document processing platform capable of understanding rubrics and contextual grading materials.
How LibreUtils achieves zero-friction quality through Bun's test runner, TypeScript pre-commit hooks, smart CI pipelines, and a developer experience designed for open-source contribution.
Exploring the philosophy, architecture, and technical decisions behind LibreUtils — an open-source collection of browser-based utilities where your data never leaves your device.
How I added scheduled publishing to a static blog using GitHub Actions and date filtering, enabling future-dated articles to appear automatically without any server-side logic or databases.
Creating a fully functional video editor that runs entirely in your browser. From timeline drag-and-drop to complex FFmpeg filter chains, here's how I built SSVE - a simple yet powerful tool for quick video edits.
Exploring the design and implementation of a web-based campaign management tool that helps Game Masters run smoother TTRPG sessions through organized file management, dynamic audio control, and real-time session tracking.
A deep dive into the evolution of the Auto-Grade project, exploring the challenges of maintaining 100% test coverage, the decision to refactor for sanity, and the performance gains from integrating GridFS.
Diving into the design and mechanics of a multi-level puzzle game inspired by Lakera's Gandalf, where players extract secrets from an LLM while navigating escalating defensive techniques and tool-based protocols.
Exploring the architecture and design decisions behind a proof-of-concept for automated bulk assignment grading using LLMs, FastAPI, and modern web technologies.
Deep dive into the performance optimizations that make albertbf.com load in under 200ms, including critical CSS inlining, edge caching, and resource optimization strategies.
How I built a high-performance blog with zero maintenance costs using Cloudflare Workers, static site generation, and modern web technologies.