Muhammad Luqman Hakim bin Syahrulnizam

FLAGSHIP

OCT 2026 // STREAMING REPLAY 12,196 SERIES

Real-Time Retail Demand Forecasting

Streamed 28 held-out days of Walmart M5 sales through Kafka into Spark Structured Streaming, landing events in a Hive-catalogued data lake and closing each business day only after control totals reconcile.
Forecast next-day demand for 12,196 SKU-store series with an MLlib random forest (backtest WAPE 0.710, live 0.725), with a weekly champion/challenger retrain that runs without blocking the stream.
Set reorder points from negative-binomial lead-time demand and a newsvendor service level; the simulated stores reached a 99.1% fill rate. Results published as a Power BI report and a live web dashboard.

Kafka Spark Structured Streaming Hive MLlib Power BI

OCT 2026 // ECONOMETRIC SIMULATION N = 109,414 GRADS

Graduate Salary & Wage Floor Model

Analysed published salary bands for 109,414 first-degree graduates (KPT Graduate Tracer Study 2024) and found 49.7% earned RM2,500 or less (Budget 2027 graduate wage floor), ranging from 27.7% in ICT to 77.2% in education.
Simulated the floor under 6 scenarios with a rank-preserving wage compression rule, estimating a 7.1–14.4% rise in the graduate wage bill.
Fitted a log-salary regression with HC3 robust errors and VIF, Breusch-Pagan, White, and RESET diagnostics on a 1,000-row calibrated sample, reproduced in Excel via LINEST and MMULT/MINVERSE.

Python statsmodels (HC3 OLS) Excel (LINEST / MMULT) Policy Simulation

OCT 2026 // STAR-SCHEMA & DAX 320 MONTHS • 51 DAX

DOSM Trade Engine

Built a 3-step Python pipeline (ingest, transform, validate) converting 3 OpenDOSM tables into star-schema fact and dimension tables covering 320 months of Malaysian trade (Jan 2000–Aug 2026) and 3,190 month-by-sector rows.
Flagged 8 export anomaly months and 25 sector spikes with z-score rules on YoY growth; validation checks halt on failure and report a 10–16% source revision gap (Jun–Jul 2024).
Modelled in Power BI with a date table and 51 DAX measures (YoY growth, 3-month moving averages, latest-month KPI cards).

Python pandas Power BI 51 DAX Measures Z-Score Anomaly

SEP 2026 – PRESENT // LIVE STREAMLIT 2X/WK AUTOMATED ETL

BudiPulse: Fuel Subsidy Tracker

Built and deployed a live dashboard tracking the weekly gap between market fuel prices and BUDI MADANI subsidised caps for RON95 and diesel, refreshed twice weekly by a scheduled GitHub Actions ETL job.
Loaded cleaned OpenDOSM price data into PostgreSQL with primary-key upserts that prevent duplicate records.
Added an interactive calculator estimating each user's monthly fuel cost against the 300 L subsidised quota.

Python PostgreSQL Streamlit GitHub Actions OpenDOSM

2024 – PRESENT // FINAL YEAR PROJECT ADJ. R² = 0.640

Smartphone Purchase Intention Study (FYP)

Built a Multiple Linear Regression (MLR) model on 100 respondents (sample size determined via G*Power) that explained 64.0% of variance in purchase intention (adj. R² = 0.640).
Identified Product Features (β = 0.431, p < 0.001) and Brand Image (β = 0.404) as the strongest predictors.
Validated full parametric assumptions including multicollinearity, homoscedasticity, and normality diagnostics.

R SPSS G*Power Multiple Linear Regression

2024 // DSC651 BI ANALYTICS 7,043 CUSTOMERS

Telco Customer Churn Dashboard (DSC651)

Analysed 7,043 IBM Telco customers in Power BI and uncovered an overall 26.54% customer churn rate.
Demonstrated that month-to-month contracts accounted for 88.55% of all churned accounts.
Isolated early-tenure vulnerability with 1,037 customers churning within their first 12 months of service.

Power BI DAX Customer Retention Cohort Analysis

MAY 2026 – OCT 2026 // FULL-STACK SQL 53 AUTOMATED TESTS

WealthWatch: Personal Finance Tracker

Built and deployed a live finance tracker serving monthly income, expense, and per-category totals from SQL aggregation queries on PostgreSQL through a 5-endpoint FastAPI service.
Engineered secure JWT sign-in and Supabase PostgreSQL Row-Level Security (RLS).
Verified backend reliability with a suite of 53 automated tests and deployed frontend on Vercel.

React FastAPI Supabase (PostgreSQL) Vercel Pytest (53 Tests)

LUQMAN HAKIM

Projects & Production Pipelines

Real-Time Retail Demand Forecasting

Graduate Salary & Wage Floor Model

DOSM Trade Engine

BudiPulse: Fuel Subsidy Tracker

Smartphone Purchase Intention Study (FYP)

Telco Customer Churn Dashboard (DSC651)

WealthWatch: Personal Finance Tracker

More Projects

Program Risk MY

PriceCatcher Retail Pricing

EV Conformal

Belanjawan Bot

Audit Fraud Detection

ESG Multi-Agent Auditor

Belanjawan 2027 Wage Cost Model

Malaysia Inclusive Economy Blueprint

Perasaan Malaysia

Antigravity GitHub Ecosystem

The Residual

Technical Telemetry, Education & Experience

Technical Stack

Experience

Education

Certifications