Volunteer Research Assistant @ The American University in Cairo

github

projects

leetcode

devto

Muhammad Azzazy • Peerlist Profile

Contributed to the development of an internal web app called the Budget Module aimed at calculating rate card for employees for the Human Capital (HC) department.
- Implemented responsive front-end components using Next.js for a smooth user experience
- Designed and integrated back-end APIs using Node.js to implement business logic
- Collaborated closely with a team of Associate Software Engineers to develop the web app

Budget Module

Led the back-end development of a Contract Wording Proof-of-Concept (PoC), focused on migrating a Power Apps solution to one based on a modern tech stack using Python Quart back-end and Next.js front-end.
- Designed, documented, and deployed APIs to Azure Functions for performing CRUD operations on Azure Cosmos DB and corresponding operations on Azure Data Lake Gen2 Storage (ADLS)
- Orchestrated the upload of contract PDFs, upserting items into Azure Cosmos DB, and triggering Azure ML pipeline for ingesting the contracts
- Engineered deletion processes for contract PDFs and associated text files, soft-deleting documents from Azure Cosmos DB container, deleting documents from gold containers in Azure Cosmos DB, and removing corresponding entries in Azure AI Search index
- Established retrieval methods for contract metadata from Azure Cosmos DB container containing contract metadata
- Derived team names from the ADLS folder structure
- Centralized and secured API access via Azure API Management
- Collaborated cross-functionally with UK-based engineers in an Agile environment via daily stand-ups and pair programming

Contract Wording Proof-of-Concept (PoC)

<h3>Overview</h3><a target="_blank" rel="noopener noreferrer" class="break-all link-break text-gray-500 hover:underline" href="https://febrilogic.com">FebriLogic</a> is a clinical decision support tool that helps healthcare professionals quickly and accurately differentiate between acute febrile illnesses (AFIs). It leverages real-world data and a novel diagnostic algorithm to provide evidence-based insights at the point of care.<h3>Features</h3>FebriLogic supports the following functionality:<ul><li>User registration and login</li><li>Password resets</li><li>Adding new patients</li><li>Updating patient information</li><li>Recording medical information for a patient including:<ul><li>negative disease tests</li><li>symptoms</li><li>biomarkers</li></ul></li><li>Determining the 3 most probable diseases from a list of acute febrile illnesses</li><li>Generating LLM-based recommendations and next steps using patient demographics, negative test results, symptoms, and biomarkers</li></ul>

FebriLogic

Led the back-end development of the Reinsurance Pricing Proof-of-Concept (PoC), focused on migrating a Streamlit app to a modern tech stack using a Python Quart back-end and Next.js front-end.
- Designed and implemented APIs for inflation factor calculations, Excel uploads to Azure Data Lake Gen2 Storage (ADLS), and claims triangle generation
- Centralized and secured API access via Azure API Management
- Collaborated with UK-based engineers through pair programming and code reviews to ensure code quality and promote cross-team knowledge transfer

Reinsurance Pricing Proof-of-Concept (PoC)

Developed time series forecasting system using SARIMAX models to predict merchant sales patterns. Implemented data extraction pipeline, performed seasonal analysis, and optimized model parameters based on performance metrics.

Demand Forecasting for Careem QUIK

Enhanced Valhalla routing engine by integrating Amman's historical traffic data to generate accurate ETAs, serving as training data for Random Forest Regression models and order pooling analysis.