TeaBlendAI
An AI-powered web application developed to modernize the tea trading industry through a digital auction system and an integrated AI chatbot. It provides role-based access, live and scheduled auction management, seller dashboards and an MSSQL-backed platform connecting tea producers with buyers.
Tea trading represents a cornerstone of Sri Lanka's agricultural export economy. Traditionally, tea lot cataloging, quality grading, and auctioning have relied heavily on manual paperwork and physical floor bidding.
TeaBlendAI was developed in collaboration with Inivos Technology to modernize this ecosystem into a digital marketplace. The platform provides structured role-based access for tea producers, registered buyers, and administrators, offering scheduled auction events, seller dashboards, and an integrated AI chatbot assistant to guide participants.
Tea estates generate discrete seasonal tea lots with distinct elevation, grade, moisture, and tasting attributes. Managing hundreds of lots across scheduled auction sessions requires:
- A systematic way for estate managers and tea sellers to catalog lots and submit reserve prices before auction deadlines.
- Clear auction scheduling and state management (drafting lots, scheduling auctions, live bidding cycles, and closing settlements).
- Role-based access boundaries to ensure tea producers control only their cataloged lots and trade reports.
- A direct, secure messaging channel between sellers and winning buyers to resolve payment, logistics, and order fulfillment.
- A robust relational database to maintain transactional consistency across all cataloged lots and auction trades.
Seller Dashboard Development
Designed and implemented the full seller dashboard user interface in Next.js, enabling tea producers to register estate details, catalog tea lots, inspect historical trade logs, and review auction timelines.
Buyer-Seller Messaging System
Architected and implemented the direct trade messaging system connecting tea sellers and buyers. Developed real-time trade conversations, negotiation channels, and order inquiry workflows with backend message persistence.
Auction Management Workflows
Built the workflows and state handling logic for scheduling tea lot auctions, configuring minimum reserve increments, and transitioning lots between scheduled and completed phases.
Backend API Design (FastAPI)
Engineered structured REST API endpoints using FastAPI in Python with strict Pydantic payload validation and role-based authentication dependencies for seller and messaging actions.
Database Modelling (MSSQL)
Designed normalized MSSQL relational schemas for estate profiles, tea lot specifications, auction schedules, order chats, and trade histories with proper foreign key constraints.
The technology stack was chosen to provide responsive client dashboards, fast asynchronous backend APIs, and enterprise relational reliability:
| LAYER | TECHNOLOGY | ENGINEERING RATIONALE |
|---|---|---|
| Frontend | Next.js, TypeScript | Provides component-driven UI architecture, efficient client routing, and responsive dashboard and chat interfaces for tea producers and buyers. |
| Backend API | FastAPI (Python) | Offers asynchronous request execution, automatic OpenAPI schema documentation, and robust request validation for auction and messaging workflows. |
| Database | Microsoft SQL Server (MSSQL) | Enterprise relational database ensuring ACID transactional integrity, relational schema enforcement, and reliable ledger and chat storage. |
| Real-Time Engine | WebSockets & REST APIs | Enables low-latency real-time bidding updates and direct buyer-seller messaging for post-auction trade negotiations. |
Click any subsystem node below to inspect modules, responsibilities, and data flows within the platform architecture:
Full-stack seller management interface enabling tea producers to onboard estates, catalog tea lots with quality attributes, manage reserves, and track scheduled auction items.
TECHNOLOGY STACKNext.js, React.js, Tailwind CSS, TypeScriptEstate onboarding, lot attribute validation, catalog workflows, seller analytics overview.
DATA FLOW & INTEGRATIONSends authenticated REST API requests to FastAPI backend with payload validation.
Browse authentic interface screenshots from the deployed platform, including multi-step auction creation wizards, scheduled auctions, live countdown biddings, historical ledgers, and the buyer-seller trade messaging system:
Engineering Challenges
Designing intuitive seller workflows for users transitioning from paper sheets required simplifying complex auction state transitions into a clear, step-by-step dashboard. Coordinating schema updates across MSSQL with FastAPI request validations ensured no unvalidated lot data entered the auction pipeline.
What I Learned
Working on an industry team project with Inivos Technology provided valuable practical experience in full-stack feature ownership—from translating domain requirements into database schemas to designing RESTful API endpoints and building responsive Next.js interfaces.