Full Stack / Mobile / Systems
SpectraLeaf
Turning tea fermentation into a readable signal.
- My role
- Software platform development · team project
- Project period
- April — September 2026
- Current stage
- Research prototype
The operational view brings live readings, batch status and fermentation controls into one dashboard.
A problem worth
connecting.
Black-tea fermentation is often assessed through manual observation. SpectraLeaf connects temperature, humidity, colour and gas readings to individual batches, giving factory teams a shared view of the process as it happens.
Built forFactory officers, factory managers and general managers
The work behind
the interface.
As part of the SpectraLeaf team at the University of Peradeniya, I designed and developed the software platform: backend APIs and data processing, the web dashboards, the React Native/Expo application, AWS and database integration, and the public project website. The wider team developed the sensing hardware and embedded system. My work also covered responsive interface design, documentation, galleries, project videos, and deployment configuration.
From readings to decisions.
A closer look.
Actual screens from the project. Select a view or open it at full size.
Factory managers review batch records, recorded quality values and pricing information.
The same system, in your hand.
Light or dark. Still connected.
Factory officer access in the light mobile interface.
The same workflow adapted to the dark mobile interface.
The public project website.
Another part of my software work: making the engineering story accessible beyond the dashboard.
How the pieces connect.
Follow an action through the system.
- 01
Sense
Readings are linked to a device, factory, batch and timestamp.
- 02
Transmit
MQTT messages enter the AWS IoT pipeline.
- 03
Process
Backend services validate readings and prepare application responses.
- 04
Store
DynamoDB holds time-series readings and batch summaries.
- 05
Understand
Web and mobile dashboards use REST APIs to display readings, history and controls.
What the platform does.
Live batch monitoring
Temperature, humidity, colour and three gas-response channels are presented alongside batch identity and fermentation status.
A view for each responsibility
Factory officers monitor operations; factory managers review batches and prices; general managers see a factory-level overview.
Fermentation controls
Start and stop actions pass through the backend and synchronise device state through AWS IoT Device Shadow.
History, quality and pricing
Historical graphs, Good Leaf Percentage recording, batch prices and revenue summaries keep batch information connected.
Mobile access
An Expo application gives factory officers access to batch monitoring and controls, with light and dark interfaces.
A public engineering record
The project website brings together the architecture, experiments, testing, hardware, AI direction, gallery and team documentation.
This is where the learning happened.
The hard part.
Keeping the sensing system, AWS services, backend, web dashboard and mobile app consistent meant handling more than a stream of numbers. Batch identity, fermentation state and incomplete cloud responses all had to agree across the platform.
How I approached it
I used a common batch-linked data structure, validated incoming readings, organised DynamoDB keys and indexes, and prepared graph-ready API responses. Device-state synchronisation and fallback handling helped the interfaces remain useful when data was incomplete or temporarily unavailable.
Tools behind the work.
- Web
- Next.jsReactTypeScriptTailwind CSSRechartsZustand
- Mobile
- React NativeExpoTypeScriptZustand
- Backend & data
- Node.jsExpressREST APIsAxiosAWS SDKDynamoDB
- Cloud & identity
- AWS IoT CoreDevice ShadowLambdaAPI GatewayAmplifyCognitoJWTGitHub Pages
Where it stands.
- A connected backend, web dashboard, mobile application and documentation platform.
- Batch-linked readings and role-based views for prototype fermentation monitoring.
- Structured data that can support future calibration and machine-learning research.
- A public technical project website deployed through GitHub Pages.