Retrieval + generationShipping

RAG knowledge systems

Private knowledge Q&A that has to cite the right doc — support, ops, and internal tools where guessing is expensive.

High-level

What I built

Designed retrieval stacks where answers must cite sources: ingest → chunk → embed → rank → generate with grounding. Aimed at support bots and internal assistants that can’t invent policy.

01Ingest docs / FAQs / policies
02Chunk + embed into vector index
03Retrieve top-k with ranking
04Generate answer with citations
05Expose via bot or dashboard

Used where hallucination is expensive: customer support, ops knowledge, and product help that has to quote the right SKU or policy line.

ContentLive

YouTube content factory

A production app for small YouTube channels: type a topic, get script, song, images, and animation prompts packaged for edit.

Script · images · song · animate prompts · package

AutomationLive

CodeBuddy

My personal WhatsApp coding buddy — allowlisted numbers, model switching, and /ask in groups so it doesn’t spam.

WhatsApp · AI chat · ACL · commands

ProductLive

Lumenden

A searchable library of 100k+ prompts and skills I built so the collection stays useful instead of a dump of folders.

100k+ prompts · skills · discovery UX