Hook
Posted Jul 18, 2026Video · 0:05by @bashifuirkashi
The video uses a high-stakes, relatable technical interview prompt to capture attention, then delivers immediate, high-value educational content in the caption to build authority and drive traffic to the creator's community.
The video presents a technical interview question about designing a RAG system for millions of PDFs, followed by a detailed breakdown in the caption. It positions the creator as an expert in AI engineering, directing viewers to their community for further learning.
When someone says “RAG over millions of PDFs” they’re not asking about AI… they’re asking about search + systems. Here’s what that actually looks like 👇 I’d break it into 5 parts: ingestion, embeddings, retrieval, generation, monitoring 1️⃣ Ingestion is offline, not request-time At scale, this must be async • Stream documents from storage (S3/GCS) • OCR only when needed • Clean + normalize text • Chunk intelligently (not randomly) • Attach rich metadata (doc, page, section, etc.) None of this should ever touch your user request path 2️⃣ Embeddings + indexing built for scale You don’t embed on demand • Batch embedding jobs (GPU or queued) • Distributed ANN indexes (Milvus, Qdrant, Vespa, Elastic) • Sharding + HNSW / IVF / PQ • Store metadata alongside vectors Key insight: metadata filtering is your first gate vector search is the fallback 3️⃣ Retrieval + generation (tight path) Your request path should stay minimal: query → metadata filter → cache → ANN search → rerank → LLM • Most queries never even hit vector search • Many don’t hit the DB at all (cache wins) • Rerank a small set only • Send 5–10 chunks max to the LLM More context ≠ better results More chunks usually hurt 4️⃣ Caching is everything This is what controls cost + latency • Query → answer cache (FAQs, repeats) • Query → retrieval cache (top chunks) • Data/index cache (hot vectors, parsed docs) Real path looks like: query → cache → (miss) retrieve + LLM → write back 5️⃣ Monitoring closes the loop Without this, your system silently degrades • Retrieval quality (recall@k) • Answer quality (feedback loops) • Latency + cache hit rates • Re-embed + re-shard as data evolves BOTTOM LINE: RAG at scale is NOT an LLM problem It’s a search + caching architecture problem Most people are building demos Real AI engineers are building systems Link in bio for the full breakdown + a group with live expert-led calls and systems for you to get hired ASAP
Interviewer: "We have millions of PDFs. How would you design a RAG system for that scale?"
Presenting a high-level technical interview question
Visualizing a developer working on the solution
Directing the viewer to the caption for the technical breakdown
Providing a structured 5-part framework for the solution
The creator's expertise and community (BASWE.Ai) are promoted via the bio link and the high-value technical content in the caption.
Read caption below and save it for later
Playing this video’s own audio, so you’ll hear the creator over the sound.
Outliers and viral hits: their videos on Overshot, all time.
AI engineering mentorship program for experienced software engineers
All time.