Dash POS Application is a mobile Point of Sale system designed for tablets, supporting complete transaction processing, sales tracking, and automatic receipt generation. The app is optimized for business environments and fully integrated with APIs for seamless data synchronization.

Payment Gateway Integration: The application utilizes the Midtrans API to handle cashless payments securely and efficiently. Cashiers can select from multiple payment options, ensuring flexibility and smooth transactions for both customers and businesses.

Secure Authentication: User sessions are managed through JWT (JSON Web Tokens), providing stateless, scalable, and secure authentication for every request. This ensures that sensitive data such as user credentials and transaction details remain protected at all times.

API Integration: The app connects to various APIs to manage and synchronize menu, order, and transaction data, maintaining consistency between the client and server in real time. It also supports multi-user access and real-time order updates across multiple devices.

I was responsible for the complete mobile development of this application, including UI implementation based on Figma designs, API integration, state management, and ensuring smooth user experience and performance optimization across devices.

Semantica is a semantic search application designed to help students find the most suitable thesis supervisors based on their research topics. The system leverages vector embedding technology and a vector database to recommend lecturers by comparing the semantic similarity between research topics and lecturers’ academic publications.

The system utilizes several embedding models such as BGE-M3, all-MiniLM-L6-v2, and IndoBERT, implemented using FlagEmbedding and SentenceTransformers. Search operations are powered by SQLite with the sqlite_vec extension and use the K-Nearest Neighbors (KNN) algorithm to identify the most relevant publication vectors for a given user query.

The system architecture consists of a Svelte (Carbon UI) front-end, a Flask API backend, and a vector search engine. All publication data is collected through web scraping from university academic repositories and stored in an optimized database structure.

Main features: paper and lecturer search, filtering by study program, embedding model selection, and adjustable result limits (Top-K). The user interface includes components like a search bar, ContentSwitcher, ExpandableTile, and Pagination for an intuitive search experience.

I was responsible for developing the web interface using Svelte and integrating the API with Flask, including implementing search endpoints and optimizing communication between the front end and the vector search system.