Query rewriting for RAG: expansion, decomposition and HyDE, explained simply

Presenting to a room
Presenting to a room

Originally published at https://pranjulrathour.scult.in/blog/query-rewriting-and-hyde-for-retrieval. That copy is the canonical version and gets updates first.

Retrieval is only as good as the query it receives, and users write queries like "that thing about refunds after 30 days for the premium plan". The document says "Refund eligibility: Premium subscribers, 30-day window". Bridging that gap before retrieval is query rewriting, and it is cheaper than most people expect.

Expansion: more words for the same meaning

Ask a small model to produce two or three alternative phrasings of the question, run retrieval on all of them, and merge with Reciprocal Rank Fusion — the same merge used in hybrid retrieval. Expansion helps most when the corpus uses formal vocabulary and users do not.

Pranjul Rathour
Pranjul Rathour

Decomposition: one question at a time

"Compare the refund policy for premium and basic plans" is two retrievals. Split it into sub-questions, retrieve for each, and give the model both contexts labelled. Without decomposition, the retriever returns whichever plan is mentioned more often and the comparison is invented.

HyDE: retrieve with a guess

Hypothetical Document Embeddings has the model write a short, plausible answer first, then embeds that answer to search with. The hypothetical text is closer in style to real passages than the question was, so it often retrieves better. The catch: if the model's guess is wrong in vocabulary, retrieval follows it off a cliff. Use it alongside the original query, never instead of it.

Pranjul Rathour, GenAI engineer, Kanpur
Pranjul Rathour, GenAI engineer, Kanpur

How to know if it helped

  1. Run your evaluation set with and without each rewrite step and compare recall@5.
  2. Log the rewritten queries in production and read a sample weekly — you will find rewrites that change the meaning.
  3. Measure added latency; an extra model call before retrieval is fine for chat, not for autocomplete.

What I actually run

For most document-QA products: hybrid retrieval on the original query plus one expansion, decomposition only when a classifier flags a multi-part question, and no HyDE by default. Simple systems with good chunking beat clever ones with bad chunks, every time.

From my carousels
7 Levels of RAG Apps
7 Levels of RAG Apps, slide 17 Levels of RAG Apps, slide 2
7 Levels of RAG Apps, slide 37 Levels of RAG Apps, slide 4
Full carousel on Instagram and LinkedIn.
Pranjul Rathour
Pranjul Rathour
GenAI engineer, Kanpur · 3x first-prize hackathon winner · campus mentor
I ship production RAG pipelines, fine-tune LLMs and build agentic AI products end to end. I lead engineering at SCULT INDIA for a 14-member team and have mentored 200+ students through TechVerse Enclave.
Open to: GenAI roles, hackathon judging, mentorship sessions and guest talks at colleges.
On stage, at hackathons and on campus
In a packed college auditorium
In a packed college auditorium
Pranjul Rathour
Pranjul Rathour
On the mic
On the mic

Comments

Post a Comment

Popular posts from this blog

Forming a hackathon team: roles, skills and the mistake most teams make

Image
At an Integral Startup Foundation hackathon Originally published at https://pranjulrathour.scult.in/blog/hackathon-team-roles-and-formation . That copy is the canonical version and gets updates first. Team TechVerse — Raman Shukla, Saksham Gupta and me — won three first prizes in 2025 not because we were the strongest coders in the room, but because we had stopped overlapping. When I mentor teams before a hackathon, formation is the first thing I fix. In a packed college auditorium The four roles Builder — owns the core feature and the demo path. Usually backend or the AI piece. Interface — owns what the judge sees: the front end, the flow, the seeded data. Storyteller — owns the problem statement, the pitch, the slides and the Q&A plan; also does user research in hour one. Integrator — keeps the pieces working together, manages the repo, runs the demo rehearsals. In a team of three, the builder or interface person takes this. The mistake Five friends who all write R...
Read more

Hello from Kanpur: what I build, and what I'll write about here

Image
On the mic Originally published at https://dev.to/pranjulrathour/hello-from-kanpur-what-i-build-and-what-ill-write-about-here-f2j . That copy is the canonical version and gets updates first. I'm Pranjul Rathour, a GenAI engineer from Kanpur, India. This is the first post on this blog, so here is who is writing, what I have actually shipped, and what you can expect if you follow along. What I build Between February and July 2026, at Next Upgrad Web Solutions, I designed, built and shipped five production AI applications end to end, alone, and deployed all of them with Docker, CI and observability: RAG.NextUpgrad : a retrieval-augmented generation service with hybrid dense plus BM25 retrieval, cross-encoder reranking, streaming cited answers, and a confidence gate that refuses to call the model when the evidence is weak. It runs in about 220 MB of memory, inside a free 512 MB instance. FineTune Studio : fine-tune open models with QLoRA in a browser. Upload a dataset, validate ...
Read more

I built 15 free tools, 1,211 prompts and a 50,000-skill library — here's what's inside tools.scult.in

Image
Carousel · 12 slides I built every part of tools.scult.in myself — 15 browser tools, 1,211 version-stamped prompts, a 50,000+ agent-skills library, a public MCP server, and 40 retro photo prompts where I'm the face in every example. All of it free. No signup, no ads, no upload step. Swipe through for what's inside and why I built it this way. Link in the last slide. Slide 1 of 12 Slide 2 of 12 Slide 3 of 12 Slide 4 of 12 Slide 5 of 12 Slide 6 of 12 Slide 7 of 12 Slide 8 of 12 Slide 9 of 12 Slide 10 of 12 Slide 11 of 12 Slide 12 of 12 Pranjul Rathour GenAI engineer, Kanpur · 3x first-prize hackathon winner · campus mentor I ship production RAG pipelines, fine-tune LLMs and build agentic AI products end to end. I lead engineering at SCULT INDIA for a 14-member team and have mentored 200+ students through TechVerse Enclave. Open to: GenAI roles, hackathon judging, mentorship sessions and guest talks at colleges. Email: pranjulrathour41@gmail.com Portfolio & ...
Read more