Fourteen years in operations — two spent turning them into AI systems
AI Solutions Designer & Digital Transformation Lead | Product Lead
Fourteen years in operations and management, and two years designing and leading the build of AI systems and agents — without writing code by hand. I design the solution, author the specs, and orchestrate delivery through AI agents and automation, owning the judgment layer: when to trust an AI output, when to reject it, and where human approval gates belong. I run a fully local AI stack — Ollama models and self-hosted n8n on Docker — so sensitive data never leaves the organization. Delivered 5 systems and 28 AI agents across real estate, HR, and government relations.
- years in operations
- 14years in operations
- building AI systems
- 2building AI systems
- systems · agents
- 5 · 28systems · agents
02Technical stack
Models
- Gemini 2.5 Flash
- Gemini 2.5 Pro
Automation
- n8nlocal
Integrations
- Slack
- Google Drive
- Google Docs
Interfaces
- React
- TypeScript
- Express
- Streamlit
- Python
Databases & Storage
- Prisma
Formats & Export
- Excel
Local Infrastructure
- Ollamalocal
- Dockerlocal
Self-hosted runtime — separate from any single system's own tools
This site
- Next.js
- React
Technical Products & Projects
Business solutions built by an operations technology lead
Appraisal Advisor — Valuation Report Review Agent
A valuation report that passes unreviewed becomes the basis for a wrong financing decision — this agent reads the report, applies 73 rules to it, and returns each finding cited to the article it rests on.
The Business Problem
Reviewing a valuation report is manual: dozens of pages checked line by line against the approved form, the executive regulation, the financing rules and IVS — so the outcome varies with whichever reviewer, and how alert they were. Errors like an empty deposit code, an all-zero adjustment grid, or a valuation date later than the issue date slip through, because catching them is tedious arithmetic rather than expert judgement.
Why We Built It
Fix the standard rather than the reviewer: one checklist applied with the same rigour every time, each finding tied to its article so it can be checked rather than believed. The split is deliberate — what arithmetic can settle stays in code, and what needs judgement goes to a model constrained to cite.
Transaction Tracking with Sequential Approvals
You always know who holds a transaction right now and where it stopped — every request moves along an ordered approval chain, and only its assigned owner can change its status.
The Business Problem
A single transaction passes through several owners and authorities before it closes, so who holds it now and where it stalled both go missing. And when no single party owns the update, the status changes from many hands and deadlines pass unnoticed.
Why We Built It
Unify transactions onto a route predefined per type — an ordered owner list passed along by approval, one update owner at any moment, and two threshold tiers that surface delay before a report asks for it. And when a particular project needs one person holding its requests, the project lead overrides the general routing without editing the routes themselves.
AuditLink — Cross-Source Audit & Saudisation Tracking
Finds whoever left and is still drawing a salary or consuming insurance — by matching four systems that know nothing of one another, through six audit rules.
The Business Problem
An employee's record is spread across four systems that know nothing of one another, so someone whose service ended keeps drawing a salary or consuming insurance for months — and it surfaces only through a manual reconciliation nobody performs every month.
Why We Built It
Bring the four sources together in one pass on one key, and turn each conflict into an action list ordered by financial exposure rather than by the order things arrived. Saudisation is measured per profession rather than across the establishment, because each profession carries its own target. And any workforce change is tried out before it is made, instead of discovered after.
Property & Collections Management Platform
Runs a whole property portfolio in one place, from the lease to the accounting entry — with 16 agents working inside it on restricted permissions and hard limits.
The Business Problem
The same figure differs from one screen to the next and nobody knows which is right. And accounting rules get broken in good faith — a sale's proceeds are added to operating revenue so the month looks excellent when it is not, and tenant deposits, which are held and returned, are counted as revenue — nobody intends the error; the tools simply never prevented it.
Why We Built It
One platform where every number is derived from its data rather than stored, with its formula declared beside it — auditable instead of merely believable. Counter-intuitive rules are printed as fixed text on the screen itself rather than left to user training, and forbidden transitions are blocked in the logic rather than in the manual. The system's vocabulary is fixed by an approved dictionary, so the same thing is never named two ways in two departments. An agent layer then sits on top, inheriting the same permissions and answering from official reports — turning an answer from a report request into a question.
Petty Cash & Task System with Two-Stage Approval
Nobody signs off on their own spending: every petty-cash custody and task passes a review then an approval by two different people, and the balance is computed from the transactions rather than stated from memory.
The Business Problem
Petty cash gets run on memory and messages: the balance is a number somebody states, invoices are scattered photos in chats, and the statement is rebuilt by hand every time it is asked for. Worse, whoever spends is whoever signs off on it — not because anyone intended that, but because nothing in the tooling prevents it.
Why We Built It
The balance should not be an opinion. Record the transaction and let the balance follow from it, so the figure becomes a trace of movements rather than a summary somebody asserts. Then move conflict-of-interest prevention out of policy and into the path: the state itself refuses to let a holder approve their own closure, so nobody has to remember the rule or confront a colleague with it. Permissions are separated at the action, not the screen — hiding a screen is user experience; refusing the action is the protection.
Smart Secretary — Agent Suite with Human Approval Gate
Twelve agents run the office's daily work — and any decision that belongs to a person stops and waits for approval before it executes, not after.
The Business Problem
Automation without oversight does what it should not; automation reviewed step by step is no longer automation. And what it did and what awaits your decision sit in a message here, a mailbox there, and a third log — so nobody sees the whole picture, and one request can be decided twice by two people who cannot see each other.
Why We Built It
One control point placed exactly where the decision lives — neither before nor after: the agent runs freely until it reaches something needing permission, then stops and holds. Around that point sits a ten-screen panel gathering everything that needs looking at — today's merged log, the approval queue, agent health, the error trail — so "what is waiting on me?" has one answer on one screen. And because the halt is real rather than advisory, the state itself carries the guard: a request is decidable only while undecided, which is what makes a second channel safe to keep.
Unified Projects Progress Dashboard
A leader sees where the whole portfolio stands at a glance — on percentages that are computed rather than estimated, and no project is declared complete before its security is tested.
The Business Problem
With projects running in parallel, “where does everything stand?” means chasing each one separately and stitching the answers together by hand. Percentages get estimated by impression, so a project is announced as nearly complete while its security has not started — because on a sheet security is a line easy to postpone, and in a slide a line easy to delete.
Why We Built It
Progress should not be a number anyone enters. Record the step and let the system compute upward to the figure leadership reads, so the indicator becomes a trace of the work rather than an estimate of it. Then move security from being a checkbox to being a phase: it counts toward the percentage like any other, it cannot be deleted from inside the interface, and it sits before handover in the order. A project whose security was never tested simply cannot reach one hundred, however much its owner would like it to.
Regulatory Qualification & Expansion Calculator
Answers the question before an expansion decision: how many branches, where, and at what cost — and computes the regulatory qualification score from seven criteria instead of estimating it.
The Business Problem
The regulatory threshold climbs over time — 40% to 50% to 60% and then 85% — and tracking it by hand across seven criteria and tiered tables is laborious and error-prone. Harder still, nobody knows in advance what a particular expansion decision does to the score, so decisions worth millions get made on an estimate rather than a calculation.
Why We Built It
Encode the regulatory methodology once as the source of truth, then build everything on top of it: the instant ruling, the improvement plan, scenario comparison, and the financial optimiser. The question shifts from "are we compliant?" to "what is the cheapest path to the next threshold?".
Professional Experience
A journey of continuous growth across HR, operations, and technology
Dar wa Emaar
Jan 2025 – Present
Public Relations Officer (PRO)
- •Managed public and governmental relations supporting organizational operations.
- •Coordinated with official authorities to ensure regulatory compliance.
- •Facilitated official correspondence, permits, and approvals.
United SEQA Group
Apr 2023 – Dec 2024
HR & GRO Supervisor
- •Led HR, GRO, payroll, and QIWA operations.
- •Oversaw recruitment, labor contracts, and employee change requests.
- •Managed Human Resources Operations.
Ibrahim Bin Sayedaan Investing Company
Jan 2020 – Apr 2023
Office Manager – GM Office
- •Managed daily operations and administrative support for the General Manager's office.
- •Coordinated meetings, schedules, correspondence, and official communications.
- •Implemented ERP MS Dynamics 365 system.
Aram Meem
May 2019 – Dec 2019
Operations HR Supervisor
- •Managed recruitment, GRO, payroll, and HR documentation across Riyadh, Jeddah, and Jordan offices.
National Technology Group
Jan 2019 – May 2019
HR Supervisor
- •Administered GOSI and HRDF processes.
- •Supported recruitment and administrative services.
Dr. Sulaiman Al Habib Medical Group
Feb 2013 – Jan 2019
Administrative Services Manager
- •Supervised administrative services and facility operations.
- •Managed cross-functional teams and improved service efficiency.
- •Government relations coordination and follow-ups with agencies.
Al Jazeera Agencies Company
Feb 2012 – Feb 2013
Government Relations Coordinator
- •Coordinated with government entities and prepared official documentation.
Education
Bachelor's Degree
Business Administration
Mid Ocean University
Core Competencies
Technology & Information Systems
Operations & Administrative Leadership
Languages
Arabic
Native
English
Professional Proficiency
Interested in the details?
Happy to talk through any system here — or a problem you're trying to solve.