KiasuCode

The Problem

Academic grade tracking is often a tedious, manual process prone to human error. Traditional tools lack engagement, making the maintenance of a consistent GPA repository feel like a chore rather than a vital part of a student's development cycle.

The Solution

KiasuCode is a production-grade Telegram bot architected with Node.js and TypeScript that acts as a student's Senior Pair Programmer. Developed through a modern AI-assisted workflow, the project highlights technical maturity by leveraging LLMs for rapid prototyping and boilerplate scaffolding while maintaining strict architectural integrity through a normalized MySQL backend, live cloud deployment, and 100% type safety.

The core innovation is a custom Dev-Lingua system that intercepts bot responses to inject a hybrid persona of Singlish and developer jargon (e.g., "LGTM", "Steady lah"), ensuring a unique and engaging user experience. The architecture ensures that the personality layer remains decoupled from the core business logic, adhering to professional software design principles.

Dev-Lingua Injection Utility

Initially designed as a global middleware, the architecture was refactored to a targeted utility function to prevent state conflicts and give absolute control over context-aware messaging.

export async function replyWithFlavor(
    ctx: Context, 
    text: string, 
    flavorType: "positive" | "negative" | "casual" = "casual"
) {
    const flavors = DEV_LINGUA_FLAVORS[flavorType];
    const randomFlavor = flavors[Math.floor(Math.random() * flavors.length)];
    
    const enhancedText = `\${text}\n\n- \${randomFlavor} -`;
    return await ctx.reply(enhancedText, { parse_mode: 'HTML' });
}