About RAG Explained Better

RAG Explained Better is a technical knowledge base about retrieval-augmented generation — how to build it, measure it, and understand why it fails. It is written by one named engineer and built on a single method: current primary sources, every number traced to a named origin, and worked examples you can check. This page says who writes it, how sources are chosen, and how any claim here can be checked or corrected.

Who writes RAG Explained Better?

Mrunmay Phanse, an AI engineer who builds retrieval-augmented generation systems. Every page carries a byline, and the author record — with the outside profiles below — is the entity behind the claims, so you can weigh who is making them:

• GitHub — the code
• LinkedIn — the professional record
• mrunmay.dev — personal site
• dev.to — writing

One author, one accountable name — not an anonymous content mill. When a page states a benchmark or a cost, it is a person putting their name to it, and the profiles above are how you check that person is real.

How are sources chosen and claims checked?

The method is the product. Every substantive page is built the same way, and the working files are kept so the claims can be traced:

• Current primary sources. Pages are built from the actual current documentation and results for the topic, so versions, prices and limits reflect what is true now — not a figure copied from an older write-up.
• Every number is sourced, never invented. A pinned library version, an embedding price, a memory formula, a benchmark figure — each traces to a named origin (a vendor’s docs, a maintained repository). Where a figure could drift, the page says to verify it before you rely on it.
• Worked examples, not assertions. A precision or recall claim is computed on a shown example so you can follow the arithmetic; a memory estimate uses the vendor’s own rule of thumb on stated inputs.
• Neutrality where it counts. On comparison pages, every tool and database is scored on the same pipeline, ordered by fit to your situation — with its real limitations printed beside its strengths.

If you want to see the method applied end to end, build a RAG pipeline is a runnable walk-through, and chunking evaluation shows the measurement side worked out in full.

Found an error? How to get it corrected.

Claims here are meant to be checkable, which means some will be wrong and should be fixed. If a number has drifted, a source has moved, or a benchmark no longer holds, reach the author through any profile in the who writes this section — GitHub is the fastest. Corrections to a factual claim are made on the page itself, not argued in a comment thread. A site that asks you to trust its numbers has to be willing to change them.