02 / PROJECT ARCHIVE
Ideas in operation. Local AI. Data engineering. Practical tools built for real-world challenges.
ALL PROJECTS AI-Assisted Financial Planning Web Data ETL & Intelligence
FILE / 01 ↗
AI-Assisted Financial Planning
Ipon Buffer — Local AI Emergency Savings Planner Sep 2026 – Oct 2026 · ASEAN Financial Health Challenge
An AI-assisted emergency-savings app that turns expense notes into user-reviewed budget entries and calculates contributions around essential expenses. Validated with 152 automated tests.
Python FastAPI React
OPEN DOSSIER + FILE / 02 ↗
Web Data ETL & Intelligence
Web Intelligence Pipeline Aug 2026 – Sep 2026 · The AI Builders Hackathon
A Python pipeline that turns public web pages into structured, AI-ready records in SQLite, with optional LLM enrichment.
Python requests BeautifulSoup
OPEN DOSSIER + 2 FILES AVAILABLE
PROJECT DOSSIER ×
AI-Assisted Financial Planning
Ipon Buffer — Local AI Emergency Savings Planner Sep 2026 – Oct 2026 · ASEAN Financial Health Challenge
Associated with Zeon Michael Group
Developed for the ASEAN Financial Health Challenge’s Emergency Savings track, Ipon Buffer helps people with irregular income plan contributions after accounting for daily essentials and upcoming bills. A locally hosted Qwen3 assistant, integrated through Ollama, converts natural-language expense notes into editable budget drafts. Users review the drafts before they affect the plan; financial calculations remain deterministic. The prototype includes exact monetary calculations, consent controls, and simulated deposits and emergency withdrawals. Validation covered 148 backend tests and 4 frontend tests, alongside browser testing. The project also includes a working demo, pitch deck, and documented AI evaluations.
Python FastAPI React TypeScript SQLite Ollama Qwen3
PROJECT DOSSIER ×
Web Data ETL & Intelligence
Web Intelligence Pipeline Aug 2026 – Sep 2026 · The AI Builders Hackathon
Associated with Athabasca University
Built for The AI Builders Hackathon, Web Intelligence Pipeline runs a public website URL through a complete workflow: Acquire → Extract → Transform → Load → Enrich. Python requests retrieves the page, BeautifulSoup extracts content and metadata, and Pydantic validates structured records. The pipeline cleans and normalizes the extracted information, then persists it in SQLite. An optional LLM layer can add structured insights such as organization, industry, products, and target customers; the core pipeline works without an LLM API key. A command-line interface demonstrates the workflow from a single URL. The implementation handles invalid URLs, HTTP errors, missing metadata, inconsistent text, and unnecessary webpage content. Future directions include multi-page collection, scheduling, stronger extraction, data-quality monitoring, and a web interface.
Python requests BeautifulSoup Pydantic SQLite LLM API