Youssef Khoury / Analyst
Youssef Khoury

Youssef Khoury · Data & Operations Analyst

I build the reports and tools my operations team runs on.

Day to day I run 40+ live orders and 20+ suppliers. I also build the Power BI, SQL and Excel reports that show what’s late and why.

↗Open live dashboard ↗See my AI agent in production ↗Download CV

Power BI · SQL Server · Excel · Power Query · DAX · Business Central · Xero

Power BI, SQL Server, DAX, Power Query, Excel, star schema and data modeling, KPI governance, requirements, process mapping, UAT, RAID logs, Business Central, Xero, dashboards, and AI agents and automation.

Selected work

2026

One system I built and run at work, plus five projects I did on my own with public and synthetic operations data. Each comes with the actual files, so you can open the workbook or dashboard and check the numbers yourself.

An AI mail agent, a team web app and a 400k+ code price search, running on real operations mail since September 2026.

Output order status, requests and follow-ups on one board, with a person approving anything that leaves the company.

40+ live orders and 20+ suppliers, with status buried in email threads, prices in dozens of spreadsheets and follow-ups in people's heads.
A Python agent reads the operations inbox every 15 minutes, a language model turns each email into structured updates, and a cloud API applies them to one team hub: requests, projects, a shipments board, stock, delivery sheets, price lookup and draft Xero invoices.
It can't send email, never guesses prices or dates, and fills exact codes only. A second AI check must agree before an order update is saved, every change can be undone in one click, and anything that could reach a client waits for review.

↗Case study + interactive demo

420-order model with a data dictionary, star schema, 8 KPIs and a live dashboard.

Output delivery risk, supplier performance and overdue cash in one weekly view, with every number traceable to its source.

420 orders across projects, suppliers, delivery milestones, invoices and payments, with no single view of where delivery or cash was at risk.
A star schema and data dictionary, 8 KPIs with written definitions, an Excel control tower, a Power BI report, and a web dashboard with an Urgent / High / Watch action queue.
On-time delivery was 69.4%. Supplier lead time drove 45.7% of delayed order value ($1.24m), the biggest single cause. $1.29m of receivables were overdue. 25 delayed orders had no recorded reason, so they’re flagged as a data gap instead of guessed.

↗Case study PDF ↗Excel workbook ↗Live dashboard

360 invoices turned into aging, a DSO proxy, promise coverage and a cash-priority queue.

Output a ranked list of which invoices to chase first, by expected cash and customer risk.

360 invoices across 48 customers, where collections chased the oldest invoices first, whatever their size or chance of being paid.
A formula-driven Excel command center with aging, overdue exposure, a DSO proxy, promise coverage and a cash-priority queue. It also has a customer risk summary, a metric dictionary and reusable SQL.
88.7% of the $949k receivables book was already overdue, and $322k of that was 90+ days old. With a 55.7-day DSO proxy, open payment promises covered only 30.5% of overdue exposure, so most of the cash at risk had no agreed pay date. Disputes, broken promises and aging are scored separately, and the queue ranks invoices by the cash most likely to come in.

↗Case study PDF ↗Excel workbook

300 purchase orders scored for spend, OTIF, fill rate, price variance, defects and supplier risk.

Output every supplier sorted into a risk tier, with the drivers behind each score.

300 purchase orders across 20 suppliers, where supplier reviews relied on anecdotes.
A scorecard rating spend, OTIF, fill rate, delay, price variance and defects. Price, delivery and quality are scored separately, then combined into one risk score with the working shown. Plus a category-spend view and reusable SQL.
OTIF ran at 37.7% while fill rate held at 98.9% and defects at 0.64%. Suppliers shipped the right goods, 4.0 days late on average. Four suppliers carrying $1.94m of the $9.61m spend scored 0% OTIF. The fix is lead times and schedules; quality was fine.

↗Case study PDF ↗Excel workbook

24 requirements with acceptance criteria, traceability, 30 UAT cases and readiness reporting.

Output every requirement traced to a test, and a go-live call that keeps failures and open items visible.

Service requests arriving with inconsistent fields, unclear ownership and unreliable SLA handling.
24 requirements with testable acceptance criteria, a requirements traceability matrix, 30 UAT cases with pass / fail / blocked evidence, a RAID log and a go-live readiness summary, all linked in one Excel workbook.
120 service requests analysed into 24 requirements, 20 of them approved at the readiness review. Every requirement is tied to a test case (100% traceability), with an 83.3% UAT pass rate: 25 pass, 3 fail, 2 blocked. 8 open high-priority items went into the go-live decision instead of being quietly closed.

↗Case study PDF ↗Excel workbook

94,480 marketplace orders analyzed across seller handoff, route, region, product category and customer-review outcomes.

Output an interactive dashboard of which sellers, routes, regions and categories drive late orders, with three practical actions.

108,449 public e-commerce item records with delivery outcomes, but no clear view of which sellers, routes, regions or categories were behind the late orders.
An order-level model that avoids double counting multi-item orders, uses single-seller orders for fair attribution, ranks sellers and routes by late orders above the network average, and keeps data-quality checks in the open.
7,543 orders arrived late. Handoff after seven days flagged a 22.8% late rate versus 5.0% within two days, but did not explain the monthly spikes; March 2018 reached 21.1% late with a 2.4-day median handoff. Low reviews reached 54.0% for late orders versus 9.1% on time.

↗Interactive dashboard ↗Method and data ↗Public dataset

Dossier index

Flagship · liveInteractive

Order value, delivery risk, supplier performance and overdue cash in one view.

The 420-order model includes a data dictionary, a SQL star schema, reusable DAX measures, eight KPIs, an Excel control tower and BA documents. It all feeds one interactive dashboard.

↗Open live dashboard ↗Excel control tower ↗Power BI file

Dataset is synthetic. Model, analysis and deliverables are original work.

● ● ●LIVE ANALYTICSFull screen ↗

Credentials

Verified
Microsoft Power BI Data Analyst Professional Certificate
Microsoft · Aug 2026

Power BI Data Analyst Professional Certificate

Data preparation, modeling, visualization, analysis and dashboard delivery.

View PDF ↗
Microsoft Business Analyst Professional Certificate
Microsoft · Jun 2026

Business Analyst Professional Certificate

Requirements, process modeling, stakeholder collaboration and solution evaluation.

View PDF ↗
Anthropic Building with the Claude API course badge
Anthropic · Oct 2026

Building with Claude API

Calling Claude from code, prompting, tool use and building AI features into workflows.

Verify ↗
Education

Bachelor of Business, Management Information Systems · Matn University College · 2020–2024

Experience

2023 – Now

Operations & Systems Specialist Alta Light

Cut end-to-end order-to-delivery lead time from 2–3 months to about 2 weeks by mapping the process, tracing recurring delays to three root causes and rebuilding supplier follow-up; management adopted 3 of 5 recommendations. Run 40+ concurrent orders and projects worth $1M+ across 20+ suppliers, and built the Excel dashboard used for weekly operations reviews. Built and run Beacon, the team’s AI operations hub: an agent reads the operations inbox every 15 minutes and turns supplier and client emails into order, shipment and request updates, with a person approving anything that goes out.

Operations Associate Alta Light

Coordinated day-to-day order-to-delivery work across suppliers, purchasing and delivery follow-up, which led to my current role.

Administrator UPM

Maintained the Excel tracker for ~30 client accounts, reported delivery and receivables status, reconciled records against invoices and payments, and owned daily client follow-up.

Financial Consultant SNA

Supported client needs assessments, underwriting submissions, policy documentation, claims, renewals and audit-ready client files.

Project Coordinator Facilitate International

Coordinated 500+ construction, maintenance and residential service requests for US and Canada clients, tracking progress, quality, issues and updates.

Contact

Ask me about any of it.

Four projects use synthetic data and one uses public data. The models, analysis and decisions are my own, and I’m happy to walk through how any of it was built.

Email youssefkhoury01@gmail.com LinkedIn /in/youssefkhoury GitHub @YoussefKhoury Download CV
Email copied