Builder Engineering Available on request

Build Your Own AI Assistant: RAG from Scratch

The assistant that actually knows your organisation's documents, because you built it on them.

Chunking, embedding, retrieval and grounding on a real corpus, including Arabic text and the traps that come with it.

Duration
6 days, 36 contact hours
Cohort size
8 to 14 participants
Delivery
In-house, Online, Blended
Languages
Arabic, English

Who it is for

Developers, technical analysts, IT teams, internal innovation units.

What people leave able to do

  • Build a working retrieval-augmented assistant grounded in your own documents
  • Chunk, embed and index a real document corpus, including Arabic
  • Design retrieval that returns the right context rather than plausible noise
  • Ground responses so the assistant says "I don't know" instead of inventing
  • Evaluate accuracy and iterate on it measurably

Modules

  1. 01 Why RAG exists
  2. 02 Embeddings explained properly
  3. 03 Chunking strategies that matter
  4. 04 Vector storage options
  5. 05 Arabic text processing and its traps
  6. 06 Retrieval quality and re-ranking
  7. 07 Grounding and refusal behaviour
  8. 08 Conversation memory
  9. 09 Evaluation and measuring accuracy
  10. 10 Deployment
  11. 11 Capstone: an assistant on your own corpus

Prerequisites

Basic Python.

Start with a short call

Fifteen minutes to understand what your team does and what you want to change. If training is not the right answer, I will say so.