Azharul Islam
Portrait of Md Azharul Islam
Sydney, NSW
Business Analyst · Master of Business Analytics

Md Azharul Islam. I work out what to build, and how sure we can be.

Business analyst with 2+ years on ERP, travel-tech and analytics platforms, now completing a Master of Business Analytics at Macquarie University. Before that, a computer science degree and co-authored research on uncertainty in neural networks.

See case studies Publications Get in touch
Monte Carlo Dropout, live the method behind my research
Hover the chart to probe
training data mean prediction ±2σ across 100 passes
A small network trains in your browser, then runs 100 times with random neurons switched off. Where the passes disagree, the model is guessing. The band widens in the gap with no data.
2+ yrsbusiness analysis, ERP and travel-tech
75.6WAM at Macquarie (Distinction avg.)
4peer-reviewed conference papers, co-author
Top 1%Vice-Chancellor's International Scholarship
Experience

Seven years on one timeline

Study, research and work overlap. Click a bar to read what each role involved.

Education

Computer science, then business analytics

An engineering foundation, followed by the business side of working with data.

Postgraduate

Master of Business Analytics

Macquarie University, Sydney · Jul 2025 – expected Jun 2027
  • Weighted Average Mark 75.6, Distinction average (as at July 2026)
  • Vice-Chancellor's International Scholarship, top 1% of international applicants
Completed units
BUSA6004 Inference, Modelling & Forecasting HD COMP6200 Data Science HD BUSA8000 Techniques in Business Analytics HD BUSA8030 Management of Data, Analytics & Change ACCG8022 Applied Business Information Systems BUSA6430 Business Applications of AI ACCG8048 Business Ethics & Sustainability COMP6010 Fundamentals of Programming
Enrolled this session
BUSA8001 Applied Predictive Analytics BUSA8090 Data & Visualisation COMP6210 Big Data MGMT8040 Project Management
Undergraduate

BSc Computer Science & Engineering

BRAC University, Dhaka · May 2019 – May 2023
  • CGPA 3.61 / 4.00 (about an Australian Distinction average), 136 credits
  • Group thesis, graded A+: uncertainty in neural network structures with Monte Carlo Dropout
  • Merit-based scholarship across three semesters
Notable grades
Data Structures A+ Statistics & Probability A+ Algorithms A Linear Algebra & Fourier A Applied Data Science A Geospatial Data Analysis A Numerical Methods A-

Strongest Macquarie units

Marks out of 100, Macquarie grade bands shaded
Case studies

How I approach product work

Client work is described without client names, commercial terms or internal documents. The analytics dashboard is my own university project and is shown in full.

Procurement risk analytics dashboard

Master of Business Analytics · led the dashboard component · Power BI

A Power BI dashboard on a simulated enterprise procurement scenario: 1,000 purchase orders migrating between legacy and target systems, where mapping failures cause invoice delays and financial exposure.

The question behind it

The goal was to tell someone what to do on Monday morning, not just to describe the data. Every view leads to an owner and an action: which business unit is breaching SLA, which projects carry the most exposure, and which root causes are worth automating first.

Four views

  • Executive: mapping success, SLA breaches by business unit, top at-risk projects by exposure
  • Operational: invoice readiness against goods-receipt ageing, rework by root cause, a filterable action list
  • Financial: exposure by business unit, risk band and rework cause
  • Automation: a what-if slider estimating SLA breaches and minutes saved as automation coverage grows
What I'd do differentlyThe dashboard explains the current state but doesn't predict. The next step is to model which purchase orders are likely to fail mapping before they do, so the tool prevents problems as well as counting them.
1,000purchase orders modelled
63.0%mapping success rate
91.2%SLA breach rate
292.88Mhigh and critical exposure

Figures are from the simulated course dataset.

Research

How certain is a neural network?

Four peer-reviewed conference papers from an undergraduate research group at BRAC University (2021–2023), led by Md Farhadul Islam and supervised by Meem Arafat Manab and Annajiat Alim Rasel. Each paper compares how different architectures express predictive uncertainty, using Monte Carlo Dropout.

My contribution

Model modification and architecture experiments, designing and revising result visualisations, solving experimental problems as they came up, and drafting and revising manuscript sections.

Research interests

Uncertainty quantification in deep learning, approximate Bayesian methods, calibration and reliability across architectures, and uncertainty-aware modelling for spatial and environmental prediction such as digital soil mapping.

Google Scholar ResearchGate
Coursework

Notebooks, code and reports

Notable assessments from my Master of Business Analytics, 2025–26. Filter by kind, and open a card for methods and results. Group work is marked, with my part stated.

Undergraduate projects

Skills

Tools, and where I've used them

Select a skill to see the work it shows up in.

Select a skill above.
Recognition

Awards and certificates

2025

Vice-Chancellor's International Scholarship

Macquarie University. Awarded to the top 1% of high-achieving international applicants.

2019–23

Merit-Based Scholarship

BRAC University. Awarded across three semesters for sustained academic performance.

2020

Best Performance Award

National Tech Carnival Traction 2020, Robotics Club of BRAC University.

Certificate

Programming for Everybody (Getting Started with Python)

University of Michigan, Coursera. Verify

Certificate

The Primary Fundamentals for a LaTeX Project

Robotics Club of BRAC University.

Contact

Let's talk.

Open to business analyst, data analyst, product and consulting roles in Sydney, and to research conversations on uncertainty quantification.

Referees available on request.