Juan J. Martinez

Computer Systems Analyst

• Evaluated and optimized the Retrieval Augmented Generation (RAG) system, improving chatbot efficiency, reducing data retrieval times, and enhancing system scalability. By optimizing relational and vector database performance, enabled seamless chatbot interactions and the efficient management of thousands of communities.
• Conducted system assessments and implemented strategic technological enhancements, leading the transition to a generative AI-powered chatbot. This initiative reduced manual content updates, automated document ingestion, and improved chatbot response accuracy by 90% through advanced prompt engineering and retrieval-based AI techniques.
• Designed, analyzed, and optimized a scalable scheduling system, performing system evaluations, database modeling, and API architecture improvements. This system streamlined tour scheduling operations, successfully supporting thousands of tours while enhancing system reliability and reducing scheduling errors.
• Led system modernization initiatives, integrating large language models (LLMs), vector search, and AI-driven automation to enhance chatbot accuracy, user engagement, and information retrieval speed. Provided data-driven recommendations that improved chatbot scalability and efficiency while supporting thousands of concurrent interactions.

  Digital Human product on a website

Software Engineer

• Developed a chatbot application for potential residents using Dialogflow SE (NLP), addressing queries about unit availability, amenities, and tour scheduling (NestJS, Typescript, Dialoglfow, PostgreSQL).
• Designed and implemented a relational database and API for storing responses, along with an admin console for conversation customization, enhancing the flexibility and scalability of the chatbot solution (NestJS, Typescript/JavaScript, PostgreSQL, Vue.js).
• Built the initial UI chatbot app with Soul Machines integration, delivering a CGI-rendered leasing agent experience and managing over 15,000 interactions monthly (HTML, CSS, Bootstrap 4, Vue.js).

The University of Texas at El Paso

Certificate in Architecting in AWS

1st Place Parkhub Challenge

Achieved first place out of 25 teams by developing a traffic prediction solution for public events, with my contribution focusing on the creation of the mobile application for end users.

ARound

Developed a mobile application that provided real-time traffic predictions and parking guidance for event attendees, contributing to a 1st Place win out of 25 teams at the Parkhub Challenge (HackDFW).

Project URL

Free Ebay Templates

Developed a no-code website enabling eBay sellers to create customizable HTML templates for product listings, attracting 250+ users and receiving positive feedback for improving listing quality and streamlining template creation.

GitHub Repository

• TypeScript
• Python
• Kotlin
• PHP
• Vue.js
• Nest.js

• Pinecone (vector database)
• Open AI API (GPT 3.5/4.0 LLM and ADA embedding model)
• Google Dialogflow SE (NLP engine)
• HTML5
• CSS
• AWS (cloud computing)
• Jenkins (CI/CD tool)
• PostgreSQL
• MySQL