I don't just build on top of abstractions — I build the abstractions. From a custom database storage engine to a fully static LMS platform, every layer of the stack has been written by hand, deliberately.
↑ BOTH LAYERS OWNED END-TO-END, ZERO FRAMEWORKS REQUIRED
A custom-built database storage engine written from scratch — implementing core data structures, disk I/O, indexing, and query resolution without relying on any existing database library.
The engine is structured as discrete, replaceable layers — from raw byte pages on disk through the indexing and query planner, up to a clean API surface. Each layer was implemented and tested independently.
Self-balancing B-tree index structure implemented from scratch, enabling O(log n) lookups and range queries over large datasets stored on disk.
Fixed-size page layout for all on-disk data. The page cache and buffer pool manage reads and writes, reducing disk I/O through intelligent eviction.
ACID-aligned transaction model — operations are atomic, consistent, and durable. Write-ahead logging ensures data is recoverable after crashes.
A query planner that decides between full table scans and index-based lookups depending on selectivity — without relying on any external query engine.
End-to-end test coverage for insert, delete, update, and range queries — verifying correctness at every layer independently before integration.
Table schema definitions with typed columns, constraints, and serialisation — all handled without an ORM, mapping directly to page-level byte layout.
The database engine and DeutschMind are built with the same ethos: own every layer, add no unnecessary dependencies, and make it run forever without maintenance overhead. One operates at the byte level. The other ships a product used by real learners. Both prove the same thing.
A fully static, browser-native German language learning platform — no npm, no bundlers, no CDN imports. 3,000+ vocabulary entries, 60+ grammar topics, and a specialised FSP module for healthcare professionals.
From absolute beginner A1 to C2 mastery — all content is CEFR-tagged and filterable without a page reload.
Every noun displays its grammatical gender visually — consistently applied across all 3,000+ entries to build intuitive gender recognition through colour association.
Text-to-speech via the Web Speech API with customisable speed, pitch, and voice selection. Zero server calls for privacy and speed. Keyboard shortcuts for hands-free study.
Dedicated Fachsprachenpüfung preparation for pharmacy and healthcare professionals seeking language certification for medical licensing in Germany. 10 domain-specific topics.
Modular vanilla JS engines for drilling vocabulary and grammar — real-time CEFR filtering, thematic grouping, and progress tracking with zero npm packages.
Pure HTML5, CSS3, and vanilla JavaScript. Loads instantly, works offline, forkable on any static host, and will keep running indefinitely without dependency rot.
Vocabulary organised into 25 thematic categories — from everyday conversation to medical and legal terminology — all filterable in real-time for targeted study.
Each topic includes a structured explanation, worked examples, and interactive exercises covering Kasus, verb conjugation, Konjunktiv, Passiv, and all major grammar structures up to C2.
A specialised section for foreign-trained pharmacists and doctors seeking Approbation in Germany. Domain-specific vocabulary and scenario training across 10 legal and professional topics.