Active Experimental Hub

GJG Labs

A specialized research playground where cutting-edge cloud architectures, autonomous artificial intelligence runtimes, and high-performance web technologies converge into functional engineering prototypes.

Omni-AI Workstation
Operational Prototype
Local-First System

Omni-AI Workstation

An Offline-First dashboard that consolidates productivity, fitness trackers, and CRM databases locally on SQLite and Drizzle ORM, connected reactively via WebSockets.

Core Objective

To develop a private, low-latency station that completely decouples personal data and task schedules from cloud services through a reactive local-first architecture.

Engine Systems
  • ⚡Offline-First local database (workstation.db) running on Bun & Drizzle ORM.
  • ⚡Granular WebSocket sync bus using 'document_change' event payloads.
  • ⚡Unified CRM timeline consolidating WhatsApp logs, Apple Notes, and task histories.
  • ⚡Life Planning module with an interactive SVG wireframe silhouette showing body metrics.
Bun.jsSQLiteDrizzle ORMWebSocketsReact
Launch Portal
Token Optimization & Matryoshka Embeddings
Completed Research
Context Engineering

Token Optimization & Matryoshka Embeddings

An interactive benchmark exploring context compression and cost reduction on LLM inputs using Matryoshka Representation Learning (MRL) and semantic retrieval filters.

Core Objective

To evaluate and prove budget optimization when retrieving semantics rather than feeding full document context into Gemini models.

Engine Systems
  • ⚡Vector dimensionality scaling from 1536, 768 down to 256 dimensions using Matryoshka Representation Learning (MRL).
  • ⚡Dual-model generation evaluation using gemini-3.5-flash and gemini-3.1-pro-preview.
  • ⚡Real-time comparison between full context (Naive) and retrieved chunks (RAG) metrics.
  • ⚡Optimized costs ($0.001140 vs $0.001824 USD on 256 dim) with minimal impact on semantic cosine retrieval (0.6735 max similarity).
Gemini APIMatryoshka EmbeddingsRAG PipelineTypeScript
Launch Portal