You will get AI and LLM features integrated into your app or product
Rising Talent
Project details
Add reliable AI features to your existing app — chat, summarization, extraction, classification, or agents — built to work in production, not just in a demo. As a startup CTO I architected all LLM integrations for a SaaS serving ~800 users at 99.7% uptime, cutting new-AI-feature validation to 24-72 hours. I'll wire the feature into your stack with structured outputs, proper error handling, prompt tuning, and cost control. Stack: Python, OpenAI/Claude, LangChain, FastAPI; React/TypeScript hooks if needed.
AI Algorithms
Large Language ModelAI Applications
Conversational AIAI Models
ChatGPTWhat's included
| Service Tiers |
Starter
$250
|
Standard
$700
|
Advanced
$1,800
|
|---|---|---|---|
| Delivery Time | 4 days | 9 days | 18 days |
Number of Revisions | 1 | 2 | 3 |
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 | - | - | - |
About Marwan
AI Engineer | LLM/RAG, AI Chatbots & Computer Vision (Python)
Jeddah, Saudi Arabia - 8:01 am local time
Over the last 3+ years I've delivered AI to real users as an engineer and startup CTO: a camera-based vital-sign (rPPG) engine running at 91% heart-rate accuracy on live video, an industrial computer-vision safety platform processing 6 video streams in real time, and production RAG pipelines answering queries in under 165 ms over 30,000+ documents. I'm also the author of 5 peer-reviewed IEEE papers and a national finalist in Saudi Arabia's Future Minerals Pioneers competition.
WHAT I CAN BUILD FOR YOU
• RAG chatbots & AI assistants over your own documents, data, or website (LangChain, FAISS/ChromaDB, OpenAI/Claude/Llama).
• LLM integrations & AI features inside your product — chat, summarization, extraction, agents, structured outputs.
• Computer vision systems — object/PPE/defect detection, tracking, real-time video analytics (YOLO v8/v11, OpenCV, RTSP, edge inference).
• AI/ML model development — training, fine-tuning, evaluation, and getting models from prototype to production.
• AI backends & APIs — FastAPI microservices, Docker, clean REST APIs, dashboards (React/TypeScript).
PROOF (real, measurable)
• rPPG vital-sign engine: 91% HR accuracy (±2.5 BPM MAE) at 28 FPS, deployed to a mobile SDK for 300+ users.
• AURA CV safety platform: 94.2% mAP PPE detection on 11,000 images; cut manual audit overhead 73%.
• Enterprise RAG pipeline: 76% precision@5, sub-165 ms latency, fully containerized.
• As CTO: shipped infra serving ~800 users at 99.7% uptime and cut new-AI-feature validation to 24–72 hours.
HOW I WORK
Clear communication, fast iteration, and honest scoping. I confirm the goal, ship a working slice quickly, and keep you updated. I write maintainable code with tests and document what I deliver so your team can run it.
I'm fluent in English and a native Arabic speaker — so I'm a strong fit for MENA/Gulf clients and for Arabic-language AI (Arabic RAG, NLP, and chatbots). Trained through SDAIA and KAUST. Available now for both one-off projects and ongoing work.
If you have an AI idea — a chatbot, an automation, a vision system, or a model — message me with what you're trying to achieve and I'll tell you exactly how I'd build it.
Steps for completing your project
After purchasing the project, send requirements so Marwan can start the project.
Delivery time starts when Marwan receives requirements from you.
Marwan works on your project following the steps below.
Revisions may occur after the delivery date.
Build & integrate your AI feature
I scope the feature, wire it into your stack with structured outputs, error handling, and prompt tuning, then deliver tested, deployable code.