Three projects, unpacked from problem to implementation. Each walkthrough follows the documented project details and makes the flow visible: the steps, decisions, and components that turn an idea into a working system.
A natural language interface for relational data. The system transforms a question into SQL through discrete, inspectable stages, using approximate matching to handle wording variation.
Users who understand their data may not know the schema or SQL syntax needed to query it directly.
SELECT and WHERE construction, multiple conditions, six comparison operators, and natural-language sorting cues.
A PDF analysis workflow that compares documents, groups related matches, and highlights shared phrases directly in annotated copies for human review.
Combining three n-gram levels helps account for phrase overlap at different scales. Multiple distance methods let users evaluate similarity in different ways.
Transitive clustering represents indirect connections too: if one document links to a second, and the second to a third, they appear in the same group.
Scores identify candidates; highlighted matching phrases help a person inspect why the documents were connected.
A desktop implementation of Baagchal, the traditional Nepali strategy game. The core rules and game state are modeled independently from the SFML presentation layer.
Tracks player turns, board state, valid movement, captures, and win-condition detection for both tigers and goats.
SFML provides the graphical board, user input, piece movement, and real-time state updates in a desktop game loop.
Object-oriented design keeps game entities, board state, movement logic, and user interaction in distinct components.