Jordana AI

Overview
Jordana AI is an innovative conversational chatbot designed to replicate the voice and personality of Jordana Green, a Minnesota radio personality and leukemia survivor. The primary goal of this project is to create a supportive digital presence for individuals affected by cancer, offering them a space to ask questions, share concerns, and seek advice from an AI that embodies Green's insights and experiences.

Objectives
Support and Information: To provide an accessible platform for patients and families affected by cancer, enabling them to engage in meaningful conversations and receive guidance on navigating their health journeys.
Legacy and Outreach: To extend Green's impact beyond her physical presence, allowing her voice and message to reach a broader audience, particularly through the National Marrow Donor Program (NMDP).
Interactive Experience: To facilitate user engagement by creating a chatbot that not only resembles Green in appearance and voice but also reflects her knowledge and personal experiences.

Development Process
Collaboration with Experts: The project was initiated by Jordana Green in partnership with Professor Manjeet Rege and his graduate students. They aimed to train the AI using transcripts from Green's radio show, providing a robust knowledge base.
Prototype Creation: Graduate student Ilyas Alhassan played a crucial role in developing an early prototype, focusing on natural language processing and user interaction design.
Ethical Considerations: Throughout the development, there was a strong emphasis on ethical implications, particularly regarding the use of Green's likeness and voice in an AI format. Rege cautioned against potential misuse but ultimately supported the project due to its meaningful purpose.

Potential Impact
Patient Support: Jordana AI could serve as a crucial resource for individuals seeking support, offering a familiar and comforting voice during challenging times.
Awareness and Advocacy: By integrating the chatbot into the NMDP website, the project aims to enhance awareness about blood cancers and encourage more people to join the donor registry.
Emotional Connection: The ability for Green's children and others facing serious health challenges to "chat" with her digital avatar provides a unique emotional connection, helping to alleviate feelings of isolation and fear.

Jordana AI Video (Click to Play)

Insurance Recommender Chatbot

Overview
The Insurance Recommender Chatbot is a project developed in collaboration with Optum, aimed at assisting users in finding suitable insurance options. This AI-driven solution focuses on streamlining the process of selecting insurance policies, making it easier for individuals to understand their choices and access personalized recommendations.

Objectives
User-Friendly Experience: To simplify the often complex and overwhelming process of selecting insurance, providing users with clear and tailored recommendations based on their specific needs and circumstances.
Accessibility: To ensure that individuals, regardless of their familiarity with insurance terms, can navigate their options effectively and make informed decisions.
Data-Driven Insights: To leverage AI algorithms to analyze user data and preferences, enabling the chatbot to provide relevant suggestions that align with users' lifestyles and financial situations.

Development Process
Collaboration with Industry Experts: The project involves partnership with Jote Taddese, a vice president of software engineering at Optum, ensuring that the chatbot is grounded in industry knowledge and expertise.
Student Involvement: Students from the University of St. Thomas are gaining hands-on experience in AI development, machine learning, and natural language processing as they work on this project, enhancing their skills and preparing them for careers in AI.
Prototyping and Testing: Similar to the Jordana AI project, initial prototypes will be developed and tested to refine the chatbot's capabilities and improve user interaction based on feedback.

Potential Impact
Informed Decision-Making: By providing clear and tailored recommendations, the chatbot could empower users to make better-informed decisions regarding their insurance options.
Increased Efficiency: The AI-driven approach can significantly reduce the time and effort required to compare different insurance policies, streamlining the process for users.
Market Adaptation: As the insurance landscape evolves, the chatbot can be updated with new policies and information, ensuring users always have access to the latest options.

Insurance Recommender Avatar Chatbot (Click to Play)

Serverless Data Pipeline for Secondhand Hounds

Overview
The Serverless Data Pipeline for Secondhand Hounds is an innovative project aimed at automating the data management processes for a non-profit animal rescue organization. The primary goal of this project is to create an automated data pipeline that captures, processes, and analyzes pet surrender applications, enabling data-driven decision-making to mitigate the rise in pet surrenders.

Objectives
Automated ETL Process: To automate the extraction, transformation, and loading (ETL) of surrender application data from the organization's website to a centralized database.
Data Integrity: To implement validation checks to ensure the accuracy and reliability of data collected during the surrender application process.
Centralized Data Storage: To utilize a cloud-based NoSQL database for efficient storage and retrieval of surrender data.
Business Intelligence Reporting: To create a dashboard that provides insights into trends related to pet surrenders, helping the organization make informed decisions.

Development Process
Collaboration with Secondhand Hounds: The project was initiated in partnership with the Secondhand Hounds team to understand their data management needs and challenges.
Cloud Service Selection: After assessing various cloud service providers, Microsoft Azure was selected for its robust offerings and non-profit support, including credits for free services.
Data Pipeline Implementation: The data pipeline was developed using serverless Azure Functions to automate the ETL process. Python was used for data transformation, and the Azure Cosmos DB was chosen for centralized storage due to its flexible schema and scalability.
Dashboard Creation: A business intelligence dashboard was built using Power BI to visualize surrender trends and facilitate data-driven decision-making.

Potential Impact
Improved Data Management: The automated data pipeline significantly reduces manual data entry, minimizing errors and enhancing data integrity.
Informed Decision-Making: With access to real-time data analytics, Secondhand Hounds can identify trends in pet surrenders and develop targeted strategies to address the root causes.
Enhanced Operational Efficiency: By streamlining data processing and reporting, the organization can allocate resources more effectively and focus on its mission of rescuing and rehoming pets.
Support for Future Growth: The scalable nature of the cloud-based solution allows Secondhand Hounds to adapt to increasing data volumes as their operations expand.

Serverless Data Pipeline for Secondhand Hounds (Click to Play)