The practice · Sydney, Australia
Riddhish.
Riddhish turns complex information into software people can use. The portfolio spans websites and desktop applications, healthcare analytics, quantitative tools and machine learning research — with the same care for the experience as the work underneath it.
From the first question to something useful.
Luna Labs is an independent Sydney practice founded by Riddhish, an actuarial studies and finance student. The work combines software engineering, machine learning and quantitative methods to solve practical problems. Sometimes the result is a model or a research paper. Often it is the application, dashboard or website that helps someone put that work to use.
The work spans the whole build: understanding the source material, choosing an approach, writing the code and shaping the interface. Healthcare projects have grown from spreadsheets and data pipelines into interactive analysis and research publishing. Alongside them, native tools, sensor-based research and workflow systems explore different ways to connect technical depth with practical use.
The name Luna comes from the moon: a reference point in a changing landscape. That idea runs through the practice: helping people find their bearings, see what matters and take the next step. The portfolio includes working products, prototypes and research; each project explains its stage and the findings that shaped it.
How the work grew
Each version leaves
something to build on.
The folders hold earlier builds, experiments and rewrites. These are a few of the threads that connect them.
Method
From the first question
to a working build.
The details change with each project. A consistent working process keeps the work focused, useful and easy to review.
-
01
Understand the decision
The starting point is the person using the work: what they need to decide, what gets in their way, and what a useful result would look like. That gives the project a clear purpose before the model or interface takes shape.
-
02
Get to know the data
An early source review examines coverage, definitions, gaps and inconsistencies. A source register, a small data audit or a working prototype helps turn assumptions into a practical scope.
-
03
Build the useful part first
The first version should answer a real question or complete a real task. The approach combines established calculations, machine learning and interface design where each makes a useful contribution.
-
04
Make the result understandable
A useful result needs context. Sources, assumptions and limitations stay within reach, and the interface lets people explore what changes the answer. Review steps matter just as much as the model behind them.
-
05
Test, refine and document
Realistic cases test the work, while a record of decisions and findings carries the learning into the next iteration. That gives the people using the result something they can maintain and question.
Capability
What that looks like
in practice.
Applied research
- Machine learning for sensor and document data
- Research designs for learning from limited labels
- Sensor fusion and trajectory estimation
- Source-led market and sector research
- Clear reporting of assumptions and uncertainty
Data and decisions
- Healthcare cost, demand and workforce models
- Interactive maps and site-screening tools
- Operating models and scenario analysis
- Quantitative strategy research and backtesting
- Dashboards that explain where a number comes from
Product engineering
- Websites and web applications
- Native desktop tools and responsive interfaces
- Data pipelines and workflow automation
- Research publishing and document generation
- Developer tools and AI integrations
Principles
Principles behind
the work.
-
01
Make it checkable
Keep the source close to the result. Someone else should be able to follow the reasoning and see where a number came from.
-
02
Be clear about the unknowns
Show which parts are measured, estimated or still being tested. That makes a model more useful to the person making the decision.
-
03
Choose the right tool
Use dependable calculations for the numbers and machine learning where it helps. The simplest approach that solves the problem is a good place to start.
-
04
Learn from what failed
Keep a record of the dead ends as well as the wins. They explain the choices in the current build and make the next version better.
Work together