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
- 01 Why RAG exists
- 02 Embeddings explained properly
- 03 Chunking strategies that matter
- 04 Vector storage options
- 05 Arabic text processing and its traps
- 06 Retrieval quality and re-ranking
- 07 Grounding and refusal behaviour
- 08 Conversation memory
- 09 Evaluation and measuring accuracy
- 10 Deployment
- 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.