Credence is an AI agentic credit assessment system built for Swin Hackathon 2026, addressing a critical gap where 70–80% of Vietnamese micro-SMEs are denied credit due to insufficient formal financial data. At its core is a LangGraph-powered agent — an 18-node stateful graph that autonomously classifies loan officer intent, plans which tools to invoke, executes multi-step ML pipelines, and streams real-time decisions back to the frontend via SSE.

The agent orchestrates five specialized tools: an XGBoost credit scoring model (AUC-ROC 0.7705, tested on 46,127 profiles), a TreeSHAP explainer, DiCE-ML counterfactual generator, FairLearn fairness validator, and a data completeness checker. Unlike LLM-only systems, Credence uses true agentic control flow — branching on intermediate results, re-assessing when data is incomplete, and combining ML outputs with RAG-based regulatory context before generating final reports. It approves 2.3× more applicants than traditional models (42.5% vs 18.6%) while achieving a lower default rate (2.13% vs 2.22%).

Erudex is an AI-powered learning platform that generates structured, pedagogically ordered lesson plans from natural language requests. Users interact with a split-view interface — a docs panel on the left for navigating hierarchical lesson content (sections → subsections) and a chat panel on the right for requesting new lessons or refining existing ones. The AI proposes a full lesson plan before any database write, letting users accept or decline before committing.

Built with Next.js 16 and the Vercel AI SDK, Erudex integrates Claude Haiku for lesson generation and supports live content enrichment via Brave Search through the Model Context Protocol. Lesson plans, sections, items, and chat history are all persisted per user in Supabase PostgreSQL using Drizzle ORM. The project was developed spec-first using Spec Kit — all features are designed, planned, and contracted in the specs/ directory before implementation.

WriteFlow is a browser extension MCP client that connects to your Google Docs to provide intelligent, context-aware document assistance. This production-scaled tool helps you work with documents faster and smarter through semantic search, natural language queries, and automatic context understanding. It features tab-aware intelligence that knows which document you're editing, RAG-powered responses using Redis vector database, and seamless Google Drive integration. Designed as both a productivity enhancer and a document intelligence platform, it provides an efficient, scalable solution for content writers, business professionals, researchers, and developers who need instant answers from their documents without manual context switching.

The Itea Lab homepage is also a modern and responsive website built with Next.js and Tailwind CSS to showcase the activities of the Itea Lab coding community. It highlights community projects, event listings, and member profiles while leveraging server-side rendering for improved performance and SEO. Designed as both an information hub and a community portal, it provides a sleek, accessible platform for developers to connect and collaborate.

SentinelAI is an AI-powered compliance automation framework designed to streamline PCI DSS v4.0 audits for financial institutions, developed as a hackathon project during the VPBank Hackathon 2025. By integrating Amazon Bedrock AI (Claude 3.5 Sonnet), AWS Config, CloudTrail, Security Hub, and Aurora PostgreSQL, it automates evidence collection, compliance evaluation, and report generation across 200+ AWS accounts.

The framework eliminates manual auditing by offering real-time dashboards, AI audit agents, cross-framework mapping, intelligent risk prioritization, and multi-format reporting. Its modular architecture includes an AI agent orchestrator, evidence collector, RAG-based knowledge base, and frontend compliance portal. With up to 90% faster evidence collection and 80% less prep time, SentinelAI demonstrates enterprise scalability, cost savings, and adaptive learning for long-term compliance management.

GoBuy is a university project that presents a modern, multilingual marketplace built with Vue.js, Express.js, Supabase, and Tailwind CSS, designed for global accessibility with real-time language switching (English, Spanish, Vietnamese) and multi-currency support (10 currencies with live FastForex API integration).

The platform provides full e-commerce functionality, including product listings with filtering, shopping cart management, checkout, and seller dashboards. It features user authentication via email/password, magic links, and Google OAuth, along with customizable user profiles.

A strong emphasis on modern UI/UX ensures a responsive, polished experience with animations, loading skeletons, and instant translations. The backend REST API supports secure JWT authentication, Supabase-managed data, and smart caching for performance.

Deployed on Vercel (frontend) and Render (backend), GoBuy demonstrates a scalable architecture for international online marketplaces, balancing accessibility, performance, and security.

Carbonio is a decentralized application for recording and verifying carbon emissions on the Solana blockchain, promoting transparency and accountability in sustainability efforts, developed as a hackathon project for BREAKOUT Hackathon by Colosseum. It enables companies to register, upload carbon data, and undergo verification by third-party auditors, while consumers can scan QR codes to view verified footprints. Built with Next.js, TypeScript, Express, and Anchor smart contracts. Carbonio also uses IPFS (Pinata) and Supabase for storage and supports NFT creation via Metaplex. By offering portals for consumers, companies, and auditors, it fosters a multi-user ecosystem that tracks and incentivizes emission reduction.