Professional record

Eid Alkhaldi

Most teams can access the same models. My work is deciding what can be trusted, what must be verified, and how AI becomes an auditable system inside regulated Arabic-English workflows.

Credential: PhD, Electrical Engineering
Base: Riyadh, Saudi Arabia
Work: Applied AI, insurance workflows, evaluation, and Arabic-English enterprise systems.

Non-commodity edge

Judgment under institutional constraints.

The durable advantage is not tool access. It is the ability to turn uncertain model behavior into evidence, controls, review points, and operating routines that a real institution can defend.

Evaluation Before Automation

I start with what must be proven: evidence recall, decision correctness, exception behavior, citation faithfulness, audit trail quality, and conditions for abstention.

Regulated Workflow Judgment

Insurance and compliance work require authority boundaries, human review, policy interpretation, exception handling, and a record of why a decision was made.

Arabic-English Enterprise Reality

Bilingual systems fail in specific ways: translated clauses drift, local terminology matters, dialect changes intent, and document hierarchy can decide the answer.

Research-Trained ML Judgment

Doctoral research and first-author publications shaped a habit of controlled comparison, reproducibility, and skepticism toward unsupported model claims.

Evidence record

Research depth, domain judgment, and bilingual deployment context.

My background combines a PhD in Electrical Engineering, first-author research in computer vision and optimization, and delivery experience across insurance, compliance-oriented automation, and bilingual Arabic-English enterprise environments.

PhD

Electrical Engineering; machine learning, optimization, and computer vision

first-author peer-reviewed AI publications

KSA

regulated enterprise AI delivery in the Saudi market

AR/EN

Arabic-English evaluation, retrieval, and workflow context

Global Insurance Innovation Forum Saudi Arabia 2025

From Buzzwords to Operational Change in Insurance

A keynote and executive panel on why insurance transformation succeeds only when data quality, change management, legacy integration, and AI controls are treated as one operating problem.

Read field note

Selected systems

Work shaped by auditability, domain constraints, and handover.

The emphasis is not demo fluency. The emphasis is whether a system remains legible when policies conflict, evidence is incomplete, and decisions carry operational consequence.

Regulated AI

Insurance AI Decision Systems

Work on underwriting, claims, fraud indicators, audit trails, and human review. The technical challenge is inseparable from policy interpretation, authority, exception handling, and operational adoption.

RAG, tools, memory

Agentic and Retrieval Architectures

Architecture for systems where retrieval, tool calls, memory, policy constraints, model evaluation, and auditability are treated as one control surface.

Bilingual enterprise context

Arabic-English AI Evaluation

Evaluation for systems that must handle Arabic and English documents, translated policy language, local terminology, dialectal intent, and evidence hierarchy.

Doctoral research

Medical Image AI Research

First-author research on ensemble optimization and histology image classification, with emphasis on reproducible experimentation and model behavior under constraint.

Essays

Field notes on AI systems that have to survive contact with institutions.

Essays on RAG, agents, MCP, insurance automation, Arabic-English evaluation, and the control systems around AI.

May 18, 2026

All essays