• 

  Home Page

• 

  Chatbot

SAM – Intelligent Immigration Assistant for H-1B & PERM

SAM is an intelligent assistant designed for international students and foreign professionals navigating the U.S. employment-based visa system.
It solves the biggest hurdles in the process — confusing rules, shifting timelines, and the challenge of finding real employers who actually sponsor — by combining live web scraping, structured database queries, and agentic AI reasoning in one unified, conversational interface.

What Makes SAM Different

• Agentic AI Orchestration
  At SAM’s core is a LangChain-powered agent that autonomously decides how to answer each query — whether to:

  • Scrape live web sources in real time
    
  • Query historical LCA/PERM data from Supabase
    
  • Combine and cross-reference multiple data sources
    
  • Perform intermediate reasoning steps before responding
    This self-directed tool selection and execution allows SAM to handle complex, multi-step immigration questions seamlessly.

• Live Data Ingestion
  Scrapes MyVisaJobs and official U.S. government websites in real time to pull employer rankings, petition counts, and wage data.

• Centralized LCA/PERM Records
  Uses a Supabase database to store and query certified Labor Condition Applications and PERM filings for historical and analytical context.

• Immigration-Aware Chatbot
  Powered by a Flask API backend and LangChain agents, SAM detects intent (e.g., “How many people from my university got H-1Bs last year?”) and executes live + database queries to return precise answers.

• Source-Linked Responses
  Every answer is paired with clickable source “bubbles” so users can verify information immediately.

Key Technical Components

Backend

• Flask REST API (/api/chat) for handling chat messages and returning structured responses
  
• Agentic LangChain core that orchestrates scraping, database queries, and reasoning steps automatically
  
• Supabase client for structured LCA/PERM queries with sync & async support
  

Live Data Layer

• src/web_sources.py scrapers for MyVisaJobs employer and university pages (FY-aware URL handling)
  
• Planned integrations with USCIS / DOL public data portals
  

Database Layer

• Supabase schema with lca_filings, lca_worksites, and perm_disclosure tables
  
• SQL JOIN queries combine filings with worksites for richer insights
  
• Async methods for AI agent compatibility
  

Frontend

• Dark-themed, modern chat UI with macOS-style chrome
  
• Message bubbles, typing indicators, quick action buttons
  
• Clickable source link bubbles for transparency
  

AI Integration

• LangChain tools for combined LCA/PERM queries
  
• Agent plans and executes multi-step workflows: scrape → query DB → filter → summarize
  
• Async support for concurrent live + static lookups
  
• Ready for LangGraph or other orchestrators for multi-agent setups
  

Impact

With SAM, visa-seekers can instantly:

• Identify top H-1B sponsoring employers
  
• See historical filing patterns for a specific company or school
  
• Compare prevailing wage data across worksites
  
• Plan their visa pathway with reliable, current information
  

Instead of juggling government XLSX files, outdated blog posts, and separate search tools, users get a single, interactive system that delivers both immediacy (live scraping) and depth (structured LCA/PERM data).

Built With

• flask
• langchain
• python
• supabase