Browse categories
Explore
Fiverr Pro
English
$
USD
I will build a rag chatbot for your documents with scanned PDF support in any language
Mahmoud Refaey
About this gig
Your team keeps asking the same questions, and the answers are buried in PDFs, policies and manuals. A chatbot only helps if it answers from those files and shows where each answer came from.
I build RAG chatbots that do that, in English, Arabic or both. Scanned PDFs are fine. OCR is where a lot of these projects quietly fail, so I treat it as real work.
Some context: at GBG I worked on most of the pipeline for an Arabic RAG chatbot over a bank's internal documents. Better OCR and retrieval helped raise answer faithfulness from 79% to 97% on our test set.
What you get:
- Your documents processed (PDF, Word, text, scans)
- Hybrid search (keyword + semantic) with reranking
- Answers with source citations
- A FastAPI backend you own, plus a simple chat page
- A short test report: what it gets right and where it struggles
Stack: Python, LangChain/LangGraph, OpenAI or Azure OpenAI (or the model you prefer), Qdrant or Azure AI Search, Docker.
Message me with what documents you have and who will use the bot. If RAG isn't the right fit, I'll tell you.
Get to know Mahmoud Refaey
Mahmoud Refaey
AI Engineer building RAG systems, LLM agents and Arabic NLP
- FromEgypt
- Member sinceNov 2024
Languages
English, Arabic
I build LLM apps that work on real, messy data: RAG chatbots over your documents, LangGraph agents, and Text-to-SQL tools that let your team query a database in plain language.
Much of my work is in Arabic. At GBG I helped take an Arabic banking RAG chatbot from 79% to 97% faithfulness, mostly by fixing OCR on scanned PDFs and improving retrieval.
Stack: Python, FastAPI, LangGraph, Azure OpenAI, Qdrant, PostgreSQL, Docker, LoRA fine-tuning.
Tell me what you're building and I'll tell you how I'd approach it.
My Portfolio
FAQ
Can it answer in different languages and read scanned PDFs?
Yes! I can build your RAG chatbot for your preferred language. Scanned PDFs can be processed using OCR to extract text. Accuracy depends on language, scan quality, and layout, so I'll test a sample first.
Will the chatbot make things up?
Any LLM can. I reduce it with good retrieval, citations, and telling the bot to say it doesn't know when your documents don't cover the question. Business and Production include a test report so you can see where it still slips.
Which AI model do you use, and who pays the API costs?
OpenAI or Azure OpenAI by default, but I can use another provider or an open-source model. Usage is billed to your own account, so you control the cost. I'll give you a rough cost per question.
Do I own the code?
Yes. You get the full source code and setup instructions. Nothing is tied to my accounts.
Can you add it to my website?
Yes, as an extra. I can give you an embeddable chat widget, or your developer can call the API from your existing site.
Is my data safe with you?
I use your documents only for your project and delete my copies after delivery. For sensitive data, everything can run in your own Azure or AWS account.
