Data Analyst & Analytics Engineer · Sydney, Australia
Computer Science graduate from UNSW Sydney, working across analytics and data engineering. I take a question from the raw extract through to the dashboard a decision is made in, so definitions stay consistent across the whole pipeline instead of being handed off midway.
Now: Data Analyst and Automation Engineer Intern at Foresight Analytics, turning LSEG and Bloomberg market data into the models the firm reports on, with the pipelines running on Azure.
Open to: Data Analyst, Analytics Engineer and graduate roles in Sydney.
Four case studies, each documented end to end. Three run on live sources; the fourth uses generated data, which is labelled as such on the project page.
| Study | What it is | Headline finding |
|---|---|---|
| Woolworths vs Coles Price Analytics | Competitor benchmarking across a year of prices from two retailers | 7.1% price parity in household goods against 62.5% in pantry — the chains compete on items shoppers can price from memory, and barely at all elsewhere |
| Australian Labour Market Dashboard | End-to-end pipeline on live ABS data | 2017 marks the point where the gender full-time gap began closing for a different underlying reason |
| SaaS Sales & Revenue Analytics | Cohort retention across 12.5K invoices | 37% of the discount-promo cohort remained at month six, against roughly 71% for neighbouring cohorts |
| YouTube Trending Analytics | Analysis of 40,000 trending videos | 38 hours is the average time a video survives on the Trending page |
woolworths-vs-coles-analytics is the fullest example of how I work, covering collection, matching, validation and reporting.
- Daily collection since July 2026, writing one immutable CSV per day. Days with no collection are recorded as missing and never backfilled with estimates.
- Entity resolution across two catalogues. National brands are matched on brand, pack size within 2% and a fuzzy name score; home brands are held separately as substitutes rather than treated as the same product. Every accepted pair stores its match score, so any pairing can be audited.
- A year of history backfilled from an open tracker keyed on the same retailer product IDs, then validated against independently collected days at 99.97% agreement over 22,616 observations before any figure was built on it.
- A pre-registered test. The item list and expected values were committed to git before any bucket-level figure was computed, so the commit timestamps establish that the predictions came first. Seven of twelve fell within range; the largest miss is analysed in detail.
- docs/data_quality.md documents four defects that reached published figures before being caught. One reversed a previously published headline, and the correction is published alongside it. The full test suite passed in all four cases.
| Project | What it is | Stack |
|---|---|---|
| Podium | Q&A platform for Atlassian townhalls, built as the UNSW COMP3900 capstone with Atlassian. Published on the Atlassian Marketplace and still in use there | Atlassian Forge, React, Node |
| Research Dashboard | Fund manager research dashboard for an investment research house, reading a SharePoint list live through the Microsoft Graph API so the research team maintains it without a developer | Next.js, Graph API, Azure SQL |
| DORA | Snowflake warehouse computing the four DORA delivery metrics, enriched in-warehouse with Cortex | Snowflake, Cortex, dbt |
| Rate My Accom NSW | Review platform for NSW student housing, with university-email verification, multi-dimensional ratings and rate limiting | Next.js, TypeScript, PostgreSQL |
| Ignite | The official PPIA UNSW platform, carrying member profiles, event tooling and a directory for the Indonesian student community | Next.js, TypeScript |
| OnlyCode | Gamified peer-to-peer coding platform with real-time collaboration, skill-based matchmaking and sandboxed execution | React, TypeScript, WebSocket |
Podium and DORA are in private repositories. Both, along with Haven and other work, are documented at melvindarialyogiana.com/projects/software.
Data
Software
Platform
- BSc Computer Science, UNSW Sydney, 2023 to 2026. GPA 3.00 / 4.00.
- Diploma in Computer Science, UNSW College, 2022 to 2023.
- Data and cloud engineering, self-directed and ongoing through DataCamp, Microsoft and Databricks.
- Previously Software Developer on Podium (UNSW COMP3900 capstone with Atlassian) and Frontend Lead at PPIA UNSW.
Awards — 1st place, CSESoc Flagship Hackathon 2025 (OnlyCode); Most Fun Idea and 3rd Best Design, UNIHACK 2026 (Peersuade), across 1,010 participants from 20+ universities; Golden Rubbish Bin, UNSW Terrible Ideas Hackathon (Stall Wars).