Overview
Skills
Job Details
Job Description:
This project will complete the development of an AI-Powered Property Tax Virtual Assistant, originally prototyped on Amazon Web Services (AWS) by the Cal Poly Digital Transformation Hub, and ensure the Virtual Assistant is maintainable. The Virtual Assistant will integrate AI, Generative AI, and semantic search to deliver accurate, authoritative answers based on official regulatory documents. It will modernize access to property tax information, reduce staff workload, and improve citizen experience.
The Virtual Assistant will help to simplify access to complex property tax regulation and exemptions forms. The Virtual Assistant will serve as a training tool for new assessor staff and reduce the workload for others. Ultimately, this will have an impact on the service provided to the taxpayers.
Key Features:
- Regular California Board of Equalization ( BOE) Updates: Vendor must ingest all new BOE updates provided to the Assessor's Office and proactively monitor the BOE website for changes (e.g., new and removed LTAs).
- Comprehensive Knowledge Base: Must include R&Ts, handbooks, technical documents, AG opinions, and related references.
- Semantic Search: Instantly finds relevant answers across thousands of pages.
- Citable Responses: Every answer includes source, page, and document reference.
- Proactive Updates: Vendor must anticipate and process information sent to the Assessor's Office promptly.
Objectives
- Develop and improve existing chatbot created by Cal Poly Digital Hub
- Understand, prepare and perform source document ingestion.
- Train the Virtual Assistant on BOE updates, regulatory texts, and guidance.
- Ensure Secure & Compliant Deployment
- Implement access controls aligned with security and compliance standards.
- Deliver a reliable AI-powered assistant tailored to property tax regulations.
- Reduce staff workload while improving transparency for citizens.
Key Activities
- Evaluation and analysis of existing prototype Complete requirements gathering, architecture design, AWS infrastructure setup.
- Build and test data ingestion pipeline, transformation scripts, integration points, and user interface
- QA, security review, documentation, training, and deployment
Core Components:
- The current prototype uses AWS technology and services (QnA Bot on AWS, Amazon Lex, Amazon Polly, and Amazon Kendra). The technology includes Natural Language Processsing (NLP) for better understanding of human questions.
- Generative AI: Leverages large language models (LLMs) and retrieval-augmented generation (RAG) for more accurate and context-aware responses.
- Integration with AWS Services: Built on Amazon Lex, Amazon Polly, Amazon OpenSearch Service, Amazon Translate, and Amazon Bedrock.
- Customizable and Scalable: Easily deployable using established templates and CI/CD pipeline tools