Spaces:
Sleeping
Sleeping
File size: 1,988 Bytes
aa63765 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 |
const steps = [
{
title: "Chunk & Embed",
detail:
"Uploads land in Celery ingestion workers, chunked to 800 chars with 100 overlap, and embedded via MiniLM or hash fallback.",
},
{
title: "Supabase / pgvector",
detail:
"Chunks upsert into tenant-scoped tables with metadata, ready for RAG MCP retrieval.",
},
{
title: "Quality Tasks",
detail:
"Nightly analytics + RAG precision@k jobs run through Celery beat (`scheduler.py`).",
},
];
export function IngestionCard() {
return (
<section className="glass-panel border border-white/10 p-6 text-white">
<div className="flex flex-col gap-4 md:flex-row md:items-start md:justify-between">
<div>
<p className="text-sm uppercase tracking-[0.5em] text-cyan-200/70">
Knowledge Ops
</p>
<h2 className="mt-2 text-3xl font-semibold">Ingestion pipeline</h2>
<p className="mt-4 text-base text-slate-300">
Drop PDFs, DOCX, MD, or direct raw text and let Celery handle the
rest. Every step mirrors the backend implementation in
`backend/workers/ingestion_worker.py`, so what you see locally is
what ships to production.
</p>
</div>
<div className="rounded-full border border-white/10 bg-white/10 px-4 py-2 text-sm text-slate-100">
Celery broker / beat ready
</div>
</div>
<ol className="mt-6 grid gap-4 md:grid-cols-3">
{steps.map((step, idx) => (
<li
key={step.title}
className="rounded-2xl border border-white/10 bg-slate-950/40 p-4"
>
<p className="text-xs uppercase tracking-[0.4em] text-slate-400">
Step {idx + 1}
</p>
<h3 className="mt-2 text-xl font-semibold">{step.title}</h3>
<p className="mt-2 text-sm text-slate-300">{step.detail}</p>
</li>
))}
</ol>
</section>
);
}
|