Khayyal
Aba‑Alkayel
I build agentic AI systems that people actually use. In hospitals, in the field, and in production.
Computer Science at Qassim University on a 4.9/5.0 GPA, and a graduate of the KAUST Academy Advanced AI Program. I led the team that won 1st place nationally at the AgentX AI Hackathon, and placed 3rd at the Saudi Digital Academy's Agentic AI bootcamp. Two of the products I work on are live with real users today, and one of them gives nurses back an hour and a half of every shift.
- GPA 4.9 / 5.0
- KAUST Academy Advanced AI
- 1st place nationally, AgentX
- 3rd place, SDA Agentic AI
- nationally, AgentX AI Hackathon
- 1stnationally, AgentX AI Hackathon
- faster clinical documentation
- 6×faster clinical documentation
- bookings on a platform I run
- 1,000+bookings on a platform I run
AgentX AI Hackathon
I led the team behind Saden: an AI agent that protects mangrove forests by tying drones, sensors and vision models together to read the environmental state of a zone and issue alerts and recommendations on its own.
- Prize
- SAR 30,000
- My role
- Team lead
- Held by
- MCIT
Held in the Qassim Region under the Ministry of Communications and Information Technology, with QABAS and T2, and judged in the presence of the Minister and the Emir of the Region.
Selected work
Click any project to read the full case study
Sadenسادن
Team Lead & AI Engineer · June 2026
A multi agent system that keeps Red Sea mangroves alive. Drones and sensors in, health verdicts and field orders out.
- of all teams, nationally
- 1stof all teams, nationally
- signals fused per zone
- 7signals fused per zone
- manual data entry steps
- 0manual data entry steps
Saalemسالم
AI Engineer & Product Manager · May 2026 to present
Clinical documentation for nurses. An AI drafts the note and checks it against hospital policy before anyone signs it.
- daily documentation time
- 2 hrs to 20 mindaily documentation time
- faster per practitioner
- 6×faster per practitioner
Wifaqوِفاق
AI Engineer & Team Lead · August 2026
A medication safety engine answering one hard question. Given everything already active in your body, is this safe right now, and if not, when?
- place, SDA hackathon
- 3rdplace, SDA hackathon
- guardrail tests
- 114guardrail tests
Enanعِنان
Product Manager · August 2026 to present
The booking platform for Saudi equestrian centres. A public marketplace plus a full operations console for every stable on it.
- equestrian centres signed
- 10+equestrian centres signed
- reservations booked
- 1,000+reservations booked
Arabic Sign Language Digital Twin
AI Engineer · In progress
Reconstructing dual hand 3D motion from ordinary video, to build a virtual smart glove without the glove.
- pose backends benchmarked
- 4pose backends benchmarked
- no sensor hardware needed
- RGB onlyno sensor hardware needed
Table Zone
Restaurant table timing SaaS. A live grid of table cards counting how long each party has been seated.
Product Manager · June 2026 to August 2026
Recognition
- 1st place nationally2026
AgentX AI Hackathon
Team lead and AI engineer for Saden. Awarded SAR 30,000, presented in the Qassim Region under the Ministry of Communications and Information Technology with QABAS and T2.
- 3rd place2026
Agentic AI Bootcamp Hackathon
Saudi Digital Academy × atomcamp. Team lead and AI engineer for Wifaq, a medication safety engine built on a deterministic rule core.
About
I design and ship AI systems end to end: the retrieval layer, the agent graph, the guardrails, and the product around them. Most of my work sits in places where being wrong is expensive: clinical documentation, medication safety, environmental monitoring. So I care more about deterministic safety rails and measurable outcomes than about model demos.
The same pattern runs through all of it: find the decision the system genuinely has to get right, make that part deterministic and testable, and let the model handle the part where language and ambiguity are the point. Wifaq's interaction verdicts come from a rule engine rather than the model for exactly that reason, which is why it still works with no API key at all.
I also run product on several of these, writing the spec, talking to the users, and deciding what not to build.
- Based in
- Saudi Arabia
- GPA
- 4.9 / 5.0
- Focus
- Agentic systems, retrieval, computer vision
- Languages
- Arabic: Native · English: Professional
Education
- 2022 to 2027, expected
BSc Computer Science
Qassim University
GPA 4.9 / 5.0 · Focus on system architecture, AI, and advanced mathematics
- 2025 to 2026
Advanced Artificial Intelligence Program
KAUST Academy
Deep learning architectures, advanced machine learning, and AI system engineering. Competitive admission, one of a limited cohort of Saudi students.
Certifications
- Advanced Artificial Intelligence Program · KAUST Academy
- Linear Algebra for Machine Learning and Data Science · DeepLearning.AI
- Calculus for Machine Learning and Data Science · DeepLearning.AI
- Probability & Statistics for Machine Learning & Data Science · DeepLearning.AI
- Python Basics · University of Michigan
- Python Functions, Files, and Dictionaries · University of Michigan
Skills
Agentic AI & LLMs
- LangGraph
- Multi-agent orchestration
- RAG
- LLM integration
- Prompt engineering
- Guardrails & evaluation
- Recommendation systems
Machine learning
- Deep learning
- Neural networks
- Computer vision
- 3D pose estimation
- Speech recognition
- Predictive modelling
Engineering
- Python
- C++
- TypeScript
- FastAPI
- React
- Postgres / Supabase
- Docker
- Git
- Pytest
Libraries & tools
- PyTorch
- TensorFlow
- Hugging Face
- OpenCV
- YOLO
- MediaPipe
- NumPy
- Pandas