Joshua – Python, AWS, LLM

The premier hospitality provider for significant, high-profile sporting events, including the Super Bowl, FIFA World Cup, and the Summer and Winter Olympic Games. Specifically for the Olympic Games, the product's web experience allows guests to choose a hotel, travel accommodations, and several events they would like to attend. It prepares a custom-built package, arranging all the bookings, ticketing, and payment.

• Architected custom blue/green deployment strategy to prepare for replacement of the legacy cart and checkout system;
• Introduced a new checkout system in time for the initial sale of Milan 2026 Winter Olympics priority access;
• Assembled and led a team for the implementation of a custom hospitality package shopping cart;
• Implemented custom integration middleware to connect the shopping cart with the B2B booking software;
• Coordinated integration of cart with e-Commerce catalog app;
• Led planning and retrospective meetings;
• Managed the schedule and expectations with a larger organization.

An early-stage prototype of LLMs to augment and process unstructured notes for sales professionals, generating shareable assets describing opportunities, key decision makers, and pitch meetings.

• Developed mobile application to interface with OpenAI APIs and custom LLMs;
• Deployed and maintained custom Large Language models;
• Implemented admin interface for training and evaluation of LLMs;
• Engineered prompts for tool calling and structured data output.

The product provides a way for users who cannot attend one of the many company-sponsored conferences to still benefit from the content of the talks provided. For a fee, users can access videos from every past conference and attend future conferences virtually.

• Developed and deployed a video ingestion pipeline;
• Added user accounts and Firebase authentication to a previously unauthenticated app;
• Enabled sharing, favoriting, and note-taking on video content;
• Synchronized account status between video library and CRM;
• Implemented video streaming for live playback.

An application designed to motivate renters to pay their rent electronically and reward them for doing so. It integrates with a wide range of property management software, giving landlords key insights into how and when renters pay their rent, and helps renters plan their finances accordingly.

• Implemented a custom data importer for Yardi data via SOAP;
• Deployed Apache Airflow to manage periodic data migration and import tasks;
• Worked with the Data Science team to design a data lake for production data;
• Oversaw and maintained multiple data pipelines to populate the DS data lake.

This country had not updated its property tax assessment system since the 1990s. As a result of accumulating issues, it had not re-evaluated property values for tax purposes in over a decade. This project aimed to unite multiple government agencies, each with its own data sets, in order to correlate the data and process it automatically using a series of AI models. The goal was to arrive at updated assessment values, which then needed to be communicated to the citizens.

• Led a team responsible for coordinating every element of the massive, multi-agency ETL pipeline core to the solution;
• Developed a web interface allowing key stakeholders insight into the running system;
• Implemented a custom document uploader;
• Integrated a third-party document lifecycle management system;
• Implemented a custom state machine and deployed it to AWS;
• Managed infrastructure for entire team (and a handful of closely aligned teams);
• Worked across teams to transition the platform from development to production.