Artificial intelligence
End-to-end AI engineering: data, training, evaluation and deployment, with models we understand and can adapt.
- Models owned and maintained in-house
- Deployment on cloud, on-premise or edge
- Arabic and English from day one
Deep-tech AI, built in-house
Our own technology, for our communities.
We research, build and deploy AI that serves Arabic and Islamic culture and understands the physical world: hadith authentication, Quran recitation analysis, Arabic OCR, AI-driven cybersecurity and computer vision in 3D.
- Flagship projects
- 4
- Service lines
- 6
- Languages: Arabic & English
- 2
01Services
What we build
End-to-end technology, from research prototype to production system, delivered by one team that understands the whole stack.
- 01
AI solutions
Custom models for detection, prediction and automation, trained on your data and deployed on cloud, on-premise or edge devices.
- Detection
- Prediction
- Automation
- 02
3D reconstruction & vision
Accurate 3D models, measurements and digital twins from cameras and video, for spaces, objects and people.
- Digital twins
- Measurement
- 3D capture
- 03
Multimodal machine learning
Systems that combine images, video, text and sensor data to reach decisions no single source can support.
- Vision + language
- Sensor fusion
- Search
- 04
Software development
Web and mobile platforms, APIs and dashboards that bring AI into daily operations, in Arabic and English.
- Web & mobile
- APIs
- Dashboards
- 05
AI consulting
Strategy, feasibility studies, data readiness and technical due diligence for organisations planning AI.
- Strategy
- Feasibility
- Data readiness
- 06
Research & development
Joint R&D, prototypes and proofs of concept with universities, government and industry partners.
- Prototypes
- Proof of concept
- Partnerships
02Projects
What we are building
Four flagship projects, led by applications that serve Islamic culture and the Arabic language.
Project 01
Hadith authentication
Supporting scholars and students in verifying hadith: analysing chains of narration, identifying narrators and matching texts across the major collections, with every result traced to its sources.
- Isnad extraction and chain visualisation
- Narrator identification and biographical cross-reference
- Matn matching across the major collections
A research tool for scholars, never a substitute for scholarly judgement.
- Isnad
- Narrators
- Arabic NLP
Project 02
Quran recitation analysis
Listening to recitation with care: recognising the surah and ayah being recited, spotting missed or changed words, and giving tajweed feedback to learners and teachers.
- Surah and ayah identification from audio
- Recitation error detection
- Tajweed feedback for learners and teachers
Built to support memorisation and teaching, alongside qualified teachers.
- Speech AI
- Recitation
- Tajweed
Project 03
Arabic OCR
Reading Arabic the way it is really written: print, handwriting and historical manuscripts of Islamic heritage, with connected letters, diacritics and mixed Arabic-English pages.
- Printed, handwritten and manuscript text
- Forms, books and archives to searchable data
- Arabic and English on the same page
- Document AI
- Manuscripts
- Diacritics
Project 04
Cybersecurity
AI that watches networks, logs and messages for anything unusual, so security teams see real threats sooner and with less noise.
- Anomaly detection on network and log data
- Phishing and fraud detection, including Arabic content
- Alert triage and security analytics
- Threat detection
- Anomalies
- Phishing
Arabic OCR · on a real manuscript
Reading a page written in 1237
Manuscripts mix calligraphy, paintings, notes in the margins and centuries of wear. Our OCR first separates the text from everything else, then finds every line and reads it.
1 · Layout
TextTextPainting2 · Lines and reading
3 · Reference transcription · colophon
- 9فرغ من نسخها العبد الفقير إلى رحمة ربه وغفرانه وعفوه
- 10يحيى بن محمود بن يحيى بن أبي الحسن بن كوريها الواسطي بخطه
- 11وصوره آخر نهار يوم السبت سادس شهر رمضان سنة أربع وثلثين
- 12وستمائة حامدا لله تعالى على نعمه ومصليا على خير خلقه سيدنا
- 13محمد النبي وآله وصحبه الأخيار الأبرار الأطهار وشرف
- 14وكرم وسلم
Point to a line to find it on the page.
An illustration of the stages of our OCR. The text shown is the manuscript's reference transcription.
Al-Hariri, Maqamat, copied and painted by Yahya al-Wasiti, Baghdad, 1237. Bibliothèque nationale de France, Arabe 5847. Source: gallica.bnf.fr / BnF.
03Platform
How we build it
A modular pipeline we own end to end, from the first camera frame to a secure system in production.
Pillar 01
Data & capture
Camera, video, document and sensor pipelines, with calibration, cleaning and labelling built in.
- Multi-camera
- Sensor data
- Annotation
Pillar 02
3D reconstruction engine
Geometry, depth and motion recovered from images: point clouds, meshes, digital twins and human pose in 3D.
- Photogrammetry
- Depth
- Pose
Pillar 03
Multimodal & Arabic language
Models that fuse vision, language and signals, including Arabic text and documents, trained and evaluated on data that reflects the region.
- Vision-language
- Arabic OCR
- Arabic NLP
Pillar 04
Secure deployment
Containerised services and APIs on cloud, on-premise or edge devices, with access control and monitoring.
- APIs
- Edge
- On-premise
04Our edge
Where we lead
Three areas of deep expertise, combined in one team and built on our own technology.
3D reconstruction
Turning ordinary images and video into precise 3D understanding of spaces, objects and human motion.
- Digital twins of buildings and sites
- Measurement and inspection from cameras
- Capture of people and movement in 3D
Multimodal machine learning
Learning from several kinds of data at once, the way people do, for more reliable decisions.
- Vision, language and sensor fusion
- Search and analytics across media
- Robust results in real conditions
Why EigenTech
- Research depth
- Ph.D.-level expertise in computer vision and machine learning leads every project.
- Our own technology
- Core components built in-house, so data and know-how stay with you and with the region.
- Bilingual by design
- Products and interfaces built for Arabic and English users from the start.
- One accountable team
- Research, engineering and delivery under one roof, from first idea to production.
05Industries
Where we work
Built for sectors where seeing, measuring and understanding the physical world makes the difference.
Government & smart cities
Construction & real estate
Healthcare
Industry & energy
Education
Islamic scholarship & heritage
Finance & banking
Security & safety
06About
Who we are
EigenTech is a deep-tech startup founded by researchers in computer vision and machine learning. "Eigen" means "own": we build our own technology, from research to production, and point it at the needs of the communities we serve.
Mission
Turn advanced research in AI and computer vision into reliable products and services that solve real problems for organisations in the region.
Vision
A region that creates its own technology: AI that is designed here, understands Arabic and English, serves Islamic culture and heritage, and keeps data and know-how close to home.
Core values
- 01
Ownership
We build our core technology ourselves, so we can explain it, maintain it and stand behind it.
- 02
Scientific rigour
Every system is measured, tested and documented to research standards before it ships.
- 03
Community first
We start from the needs of local people and institutions, not from the latest trend.
- 04
Responsible AI
Privacy, fairness and transparency are design requirements, not afterthoughts.
Leadership
Led by a founder with doctoral-level expertise in computer engineering and computer vision.
Dr. Ahmed Elhayek
Founder & Chief Executive Officer
Ph.D. in Computer Engineering with deep expertise in computer vision, 3D reconstruction and machine learning. Leads EigenTech’s research, product strategy and partnerships.
- Ph.D. Computer Engineering
- Computer vision
- 3D reconstruction
- Machine learning
07Contact
Let’s build what’s next
Whether you need an AI system built from scratch or want to bring vision and intelligence into what you already run, tell us about it. We reply within two working days.