Arun Kulkarni's Resume

Professional Experience

Software Development Engineer

University System of Maryland Office

Jan 2023 - Present

Replaced manual paper processes for ~175 retirees with a secure one-time pad-based “magic link” login (64-bit tokens, 10-minute validity) in PHP/SQL Server, meeting FERPA standards and reducing support overhead by giving HR admins self-service tools.
Transitioned from live production edits to a Git-based workflow with multi-stage CI/CD pipelines in Azure DevOps, cutting deployment times by ~66% and enabling path-based triggers in a large monorepo for more efficient, targeted releases.
Modernized the budget application system by updating ~15 critical financial reports for state compliance and implementing bulk-insert T-SQL loaders, resulting in a 10-30x speed increase and planned self-service access for ~100 users across 12 institutions.
Migrated 200+ sensitive Board of Regents PDFs to SharePoint, eliminating a vulnerable “security-by-obscurity” approach and building a custom interface that renders files on-demand post-authentication to improve data governance.
Launched daily 10-15 minute stand-ups for a 4-person dev team to replace infrequent weekly meetings, improving task visibility, accountability, and overall communication.
Upgraded 10 legacy systems from PHP 7.4 to 8.2.x—including phpMyAdmin and SimpleSAMLphp—to address compliance audits and security gaps, ensuring continued vendor support.
Began developing a containerized FastAPI backend for internal GenAI features, moving toward a microservice-oriented architecture that standardizes local and production environments via Docker.

Software Engineer I

Meso Scale Diagnostics

Mar 2022 - Jan 2023

Designed end-to-end React/GraphQL applications, reducing API calls by ~30% and streamlining manufacturing workflows under strict regulatory requirements.
Built token-based authentication and role-based access controls in .NET (MVC/MVVM), collaborating with QA to ensure compliance, reliability, and production-ready solutions.
Introduced Cypress for automated end-to-end testing, lowering manual testing overhead and improving release quality.
Gathered and refined business requirements with cross-functional teams, delivering stable, scalable applications that addressed manufacturing, compliance, and IT demands.

Software Engineer Intern → Associate → Software Engineer I

Bridgeway Benefit Technologies

Jan 2021 - Mar 2022

Progressed rapidly from Intern to Engineer, contributing to .NET-based enterprise solutions that integrated and modernized health claims data workflows.
Developed an internal MS Teams app with C# (Blazor), Azure Functions, and MS Graph API to automate ~100 monthly ticket requests; deployed to Azure App Service and led user workshops.
Built .NET desktop applications, including Excel-to-database importers and single sign-on solutions using Azure Key Vault, improving data accuracy and security.
Restored post-security event build pipelines by migrating to Continua CI, strengthening deployment reliability and response times.

Selected Projects

Click-N-Ship Modernization (MBA Capstone)

Led strategic planning for a USPS modernization effort, conducting user surveys, competitor analyses, and stakeholder engagements. Proposed improved UI/UX, AI chatbot integration, and analytics enhancements.

Agentive AI Teacher Assistant (In Progress)

Building an AI-driven assistant using PydanticAI, Retrieval-Augmented Generation (RAG), and OpenAI to automate lesson planning, homework generation, and multilingual parent communications.

Cloud Resume Challenge (This Page)

Constructed a serverless interactive resume hosted on Azure Static Web Apps, integrated analytics with Azure Functions/CosmosDB, and employed GitHub Actions for automated CI/CD.

View Project Source: Frontend | Backend

Regression, Clustering, Classification of COVID-19 in Maryland

Developed a Python-based data pipeline to analyze ~449 joined records from the Maryland Department of Health and 2010 Census data. Implemented linear regression (R² > 0.9) to correlate population metrics with COVID-19 cases, then used k-means clustering and classification algorithms (KNN, decision trees, naive Bayes, random forest) to predict case severity at the zip-code level. Achieved 90–96% accuracy and demonstrated the feasibility of using demographic and case data to model public-health insights.

Arun Kulkarni

Professional Summary

Technical Skills

Certifications