AI & Data · RP-AI-0007

Rayen Fekih

SeniorAI Engineer (Generative AI)

GenAI engineer for Arabic semantic search and multi-agent RAG on GCP, plus production vision and pricing ML

  • Identity verified
  • Remote Pro Ai team member · under contract

Availability on request

  • 5yearsprofessional experience
  • SeniorSeniority level
  • Agentic SystemsPrimary stack
  • QatarBased in · GMT+3

Rayen Fekih

Profile summary

AI/ML engineer with 5 years of production experience across generative AI, Arabic NLP, computer vision and data science. At iHorizons in Qatar, cut semantic-search latency from 12s to 4s across 165K+ Arabic fatwas and built a Google ADK multi-agent sermon assistant adopted across 1,000+ mosques. Previously engineered LangGraph agent harnesses, dynamic pricing and recommendation systems at Infor, and retail forecasting over 10M+ transactions at Cognira. Google Cloud Certified Professional Machine Learning Engineer.

  • Languages

    • Arabic
    • English
    • French
  • Sectors

    • Government
    • Energy
    • Retail
    • SaaS

Rayen Fekih

3 roles

Experience

Most recent role first, with outcomes as stated in the CV.

  1. AI Engineer

    iHorizons

    – PresentQatar

    • Reduced end-to-end semantic-search latency from 12s to 4s on the IslamWeb platform by optimising retrieval architecture, vector search and BigQuery execution.
    • Improved Arabic semantic search across IslamWeb’s 165K+ fatwas with hybrid retrieval combining Gemini embeddings, BigQuery Vector Search, BM25 and RRF.
    • Built a multi-agent Khateeb assistant for Qatar’s Ministry of Islamic Affairs, supporting adoption across 1,000+ mosques with grounded RAG and an ADK agent harness for planning, retrieval, generation, evaluation, moderation and revision.
    • Automated daily IslamWeb content processing on Python/GCP pipelines for embeddings, Arabic summarisation, classification, Quran-aware TTS and search-index refresh.
    • Delivered oil-rig CCTV safety monitoring that detects PPE violations and restricted-zone entry, fine-tuning YOLO and RF-DETR with OpenCV multi-object tracking.
  2. Machine Learning Engineer

    Infor

    – New York, USA

    • Contributed to a 5% profit increase for an enterprise customer with dynamic pricing models personalised to purchasing behaviour.
    • Reduced recommendation response time by 80% by optimising upsell/cross-sell inference with NumPy and Cython.
    • Built a reusable LangGraph agent harness with tool calling, a Qdrant RAG store and contextual memory for reuse across customer implementations.
    • Automated build and deployment of Docker-containerised AI services through CI/CD pipelines.
  3. Data Scientist

    Cognira

    – Tunisia

    • Contributed to a 15% profit-margin improvement for a cheese manufacturer with regression and explainability models that optimised process parameters.
    • Built a retail demand-forecasting product for sales prediction, promotion planning and what-if analysis, on PySpark, SQL and time-series pipelines over 10M+ transactions.

Rayen Fekih

3 projects

Selected projects

Shipped titles and products, with the role played on each.

  • Multi-Agent Khateeb Assistant

    AI engineer

    Grounded sermon-writing assistant drawing on Quran, Hadith and trusted Islamic books under strict content constraints, adopted across 1,000+ mosques in Qatar.

    Built with

    • Google ADK
    • Agent Engine
    • Gemini
    • RAG
  • IslamWeb Arabic Semantic Search Platform

    AI engineer

    Hybrid Arabic semantic search over 165K+ fatwas with latency reduced from 12s to 4s.

    Built with

    • Gemini embeddings
    • BigQuery Vector Search
    • BM25
    • GCP
  • Computer Vision Oil Rig Safety Monitoring

    Computer vision engineer

    CCTV application detecting PPE violations and restricted-zone entry for automated worker-safety monitoring.

    Built with

    • PyTorch
    • YOLO
    • RF-DETR
    • Roboflow
    • OpenCV
    • MLflow

Rayen Fekih

8 core skills

Skills

Core skills first, then the full toolkit by theme.

Core skills

  • Agentic Systems
  • RAG & Semantic Search
  • Arabic NLP
  • LangGraph
  • Google ADK
  • Vertex AI / GCP
  • PyTorch
  • LLM Evaluation
  • Generative AI & NLP

    • Agentic systems
    • RAG
    • LLM evaluation
    • Semantic search
    • Hybrid retrieval (BM25, RRF)
    • Arabic NLP
    • LangGraph
    • Google ADK
    • Gemini
  • Computer vision

    • PyTorch
    • YOLO
    • RF-DETR
    • OpenCV
    • Multi-object tracking
    • Roboflow
  • Machine learning & data

    • Python
    • scikit-learn
    • Recommendation systems
    • Forecasting
    • Clustering
    • PySpark
    • SQL
    • BigQuery
    • Cython
  • Cloud & MLOps

    • GCP
    • Vertex AI
    • Agent Engine
    • Qdrant
    • Docker
    • FastAPI
    • CI/CD
    • MLflow

Rayen Fekih

Education and certifications

Degrees, certifications and awards on record.

Education

  • Bachelor’s Degree in Computer Science – Artificial Intelligence SpecializationNational Engineering School of Tunis (ENIT) · 2021

Certifications

  • Google Cloud Certified Professional Machine Learning EngineerGoogle Cloud
  • Claude Certified Architect, Foundations (CCAR-F)

Rayen Fekih

Usually works with

Roles this profile is usually paired with in a squad.

AI & Data

Bring this profile into your squad

Add the role and level to the Squad Builder; availability is confirmed with your quote.

Availability on request