zmazz.dev

Project

General retrieval

Prototype · Paused · 2025

Capability-level case study

LLM- and cloud-agnostic knowledge-base structurer intended to work across different verticals without a single-vendor retrieval stack.

Role
Author
Context
Personal
Capabilities
Knowledge architecture, Document ingestion, Hybrid retrieval

Overview

Retrieval stacks rot when they marry one model vendor and one cloud. This prototype is a knowledge-base structurer meant to stay LLM- and cloud-agnostic so the same bones can serve different verticals.

I am building it alone. Source: GitHub description only.

Current state

Prototype, last updated late 2025. The source repository is private and is not linked from this page.

Limitations

“Agnostic” is a design constraint, not a completed portability proof. No benchmarks and no claim that this replaced the professional retrieval work.