Cuevo brings curriculum, learning activity, assessment evidence and feedback into a connected experience for learners, educators and institutions. The product is being developed in English and Arabic, with support for right-to-left experiences. My work at E Deviser connects product direction with engineering and agentic intelligence, including evidence-grounded recommendations for what a learner or class should do next.
TOOLS & TECHNOLOGY
RAG workflowsScoped retrieval and relevant context
Retrieval and agentic intelligence form part of my product and engineering work on Cuevo at E Deviser.
02 / MY CONTRIBUTION
Product direction and agentic intelligence.
01
Product direction
I lead product strategy for Cuevo, shaping connected learning workflows and personalized next steps for learners, educators and institutions.
02
Connected education workflows
I contribute to architecture and development for the needs of educators, learners and institutions, with Arabic and English experiences.
03
Agentic intelligence
My work includes agentic and retrieval-augmented systems with controlled autonomy, alongside testing, security and observability practices.
03 / THE WORKFLOW
From context to meaningful action.
01
STAGE 1 OF 5
Understand the learning context
Identify the learner or class, the curriculum and the question that needs attention.
02
STAGE 2 OF 5
Gather relevant evidence
Bring permitted assessment, activity and feedback evidence into the request.
03
STAGE 3 OF 5
Reason about next steps
Connect the evidence with a recommendation, rationale and uncertainty.
04
STAGE 4 OF 5
Review meaningful action
Keep consequential decisions subject to educator or human approval.
05
STAGE 5 OF 5
Learn from feedback
Use subsequent activity and feedback to inform the next learning decision.
PROJECT IN MOTIONLearning, in motion.▷AI-generated concept animation of light moving through a glass book. A visual metaphor for learning, created with Sora 2.04 / DESIGN CONSIDERATIONS
Personalization, control and trust.
Relevant evidence
A useful learning recommendation needs the right curriculum and learner context. Retrieval should preserve access boundaries and make the evidence understandable.
Educator control
A recommendation and authority to act are separate concerns. Consequential educational decisions need a clear human review boundary.
Clarity across languages
English and Arabic experiences need consistent meaning, readable right-to-left layouts and usable learning workflows.
↳
Cuevo is the product; E Deviser is the company. The system sequence summarizes agentic learning workflows and is not a deployment diagram. The cover is original concept art; the considerations describe design tensions rather than claim particular released behavior.
