Visa Movement Tracker

Visa Movement Tracker Using Hybrid Machine Learning Model: Empowering U.S. Green Card Applicants with Predictive Insights and AI Assistance

Comment 1

Inspiration

The U.S. green card process is filled with uncertainty. The Visa Bulletin, while crucial, is confusing, hard to compare month to month, and offers no future outlook. As immigrants ourselves, we wanted to build something that reduces that uncertainty and helps people plan major life decisions—careers, education, housing, and family—with more confidence.

What It Does

Visa Movement makes the Visa Bulletin accessible and predictive. It provides:

  1. Forecasts of future cutoff dates using machine learning
  2. An interactive Visa Table that auto-updates every month
  3. An AI-powered assistant that answers immigration questions in plain language with context from the latest data

How We Built It

  1. Collected and structured historical Visa Bulletin data into JSON time series
  2. Combined ARIMA for short-term stability with Prophet for long-term trends and seasonal effects
  3. Automated bulletin updates with a GitHub Action and JSONBin API, so new data is available instantly
  4. Integrated a ChatGPT-based assistant to provide conversational explanations and guidance

Challenges We Ran Into

  1. Normalizing and structuring inconsistent government Visa Bulletin data into a usable format
  2. Designing a forecasting model that balances accuracy with conservative, trustworthy predictions
  3. Implementing secure API calls by moving requests to a backend server and configuring CORS
  4. Maintaining accuracy and reliability in the AI-powered assistant
  5. Managing API usage efficiently to avoid unnecessary costs with OpenAI

Accomplishments That We’re Proud Of

  1. Built the first open-source web tracker that forecasts Visa Bulletin movements
  2. Automated the entire data pipeline—from scraping to live frontend updates
  3. Seamlessly integrated machine learning predictions with a natural language chatbot

What We Learned

  1. How to work with APIs securely using backend servers and environment variables
  2. Combining ARIMA and Prophet models allowed us to build a more robust forecasting system that captures both short-term trends and long-term, irregular shifts in visa bulletin movements
  3. The importance of clean data contracts between backend and frontend
  4. Automating data collection and updates with CRON jobs and JSONBin
  5. Building a usable chatbot in React with state management and multi-turn memory

What’s Next for Visa Movement

  1. Personalized user accounts to track individual priority dates
  2. Alerts via email/SMS when cutoff dates move
  3. Improving the accuracy and reliability of the AI-powered chatbot
  4. Integration of USCIS processing times and policy-change detection Expansion to cover other visa categories

Built With

Share this project:

Updates