You will get a custom AI chatbot trained on your business documents
Rising Talent
Rising Talent
Project details
Most AI chatbots fail in business for one reason: they answer confidently when they don't actually know. I build assistants that do the opposite — every answer cites the exact document and page it came from, and when your documents don't contain the answer, the system says so instead of inventing one.
I work as a GenAI engineer for a large international travel-tech company, building agent and retrieval systems that run in production on Kubernetes and Azure. The same engineering practice goes into every project here: structured parsing that respects tables and clauses, hybrid retrieval, and accuracy measured against a real test set — not a demo that works once.
Before I write any code, I test your questions against your documents and tell you honestly whether this will work. If your files won't parse well or your answers live in people's heads rather than in writing, I'd rather say that on day one than deliver something disappointing on day ten.
You receive the source code, the deployment guide and the credentials. Nothing is locked to me, and nothing is locked to a single AI provider — you can swap models later without a rebuild.
I work as a GenAI engineer for a large international travel-tech company, building agent and retrieval systems that run in production on Kubernetes and Azure. The same engineering practice goes into every project here: structured parsing that respects tables and clauses, hybrid retrieval, and accuracy measured against a real test set — not a demo that works once.
Before I write any code, I test your questions against your documents and tell you honestly whether this will work. If your files won't parse well or your answers live in people's heads rather than in writing, I'd rather say that on day one than deliver something disappointing on day ten.
You receive the source code, the deployment guide and the credentials. Nothing is locked to me, and nothing is locked to a single AI provider — you can swap models later without a rebuild.
AI Algorithms
Large Language Model, Transformer ModelAI Applications
AI Chatbot, AI-Enhanced Classification, Conversational AI, Natural Language Generation, Natural Language Understanding, Text RecognitionAI Development Language
PythonAI Tools
Azure OpenAI, GitHub Copilot, Gradio, Hugging Face, PyTorch, StreamlitAI Models
BERT, ChatGPT, GPT-3, GPT-4, LLaMAWhat's included
| Service Tiers |
Starter
$180
|
Standard
$450
|
Advanced
$1,100
|
|---|---|---|---|
| Delivery Time | 5 days | 10 days | 20 days |
Number of Revisions | 1 | 3 | 5 |
AI Model Integration | |||
Batch Normalization | - | - | - |
Database Integration | - | ||
Detailed Code Comments | - | ||
Image Upscaling | - | - | - |
MLOps | - | - | |
Model Deployment | - | ||
Model Documentation | - | ||
Model Monitoring | - | - | |
Model Testing & Optimization | - | ||
Model Tuning | - | - | |
Natural Language Processing | |||
NLP Tokenization | |||
Pre-Training | - | - | - |
Prompt Engineering | |||
Setup File | |||
Source Code |
Frequently asked questions
About Mohamed
LLM/AI Engineer | AI Automation, AI Agents | Python, RAG, NLP
Nice Saint-Augustin, France - 2:05 pm local time
Most companies have the answers they need — buried in PDFs, contracts, handbooks and shared drives nobody has time to search. I build AI assistants that read those documents and answer questions about them in plain language, citing the exact page every time. When the answer isn't in your files, the system says so instead of making one up.
I work as an AI engineer for a large international travel-tech company, where I build assistants and automations that run in production for internal teams. On Upwork I take scoped, fixed-price projects, so you know the deliverable and the price before we start.
What I build
Document assistants — ask questions across your contracts, policies, reports or knowledge base and get sourced answers in seconds instead of an afternoon of searching
AI agents for a bounded job — reading and sorting inbound emails and PDFs, extracting fields into your CRM or spreadsheet, drafting first-pass replies with a human approving before anything is sent
Automation of manual work — the copy-paste between systems, the recurring report someone rebuilds every Monday
Rescue work — an AI prototype that gives wrong answers, costs too much to run, or worked in the demo and not in reality
How I work
Fixed price, agreed before I start. No hourly surprises.
I test your questions against your documents first and tell you honestly whether this will work. If your answers live in people's heads rather than in writing, I'd rather say so on day one.
Every project ships with the source code, documentation and credentials. Nothing is locked to me or to a single AI provider — you can switch models later without a rebuild.
Your data can stay entirely inside your own infrastructure if your industry requires it.
Tools I work with: Python, Azure OpenAI, Claude, open-weight models (Llama, Mistral), LangChain/LangGraph, PostgreSQL, Docker, Kubernetes.
Send me a sample of your documents and the questions you wish they could answer, and I'll tell you where I'd start.
Steps for completing your project
After purchasing the project, send requirements so Mohamed can start the project.
Delivery time starts when Mohamed receives requirements from you.
Mohamed works on your project following the steps below.
Revisions may occur after the delivery date.
Scoping call and document review
A 20–30 minute call to confirm scope, then I inspect your files to check parsing quality and flag anything problematic. You get a written summary with the architecture and confirmed deliverables before any code is written.
Ingestion pipeline and indexing
I build the parsing pipeline for your formats, including OCR for scans, split documents into structured passages, and index them with their exact source location so every future answer can cite a page.