Ian Harrison

Sterling Seacrest Pritchard · Atlanta, GA

• Lead and assist security response for company security events, including phishing investigation and remediation.
• Own the management and security of all information in the company data lake, deployment of Cloud Run services, and integration with third-party applications.
• Head of AI integration — lead bi-weekly best-practice sessions for 100 users and serve in an advisory role for Executive Committee meetings.
• Advise on and implement modifications to the company website, ensuring compliance standards are maintained.
• Developed internal systems for company-wide budget calculation, information collection, and process automation.

Health Industries LLC · Birmingham, AL

• Authored documentation and testing procedures for security protocols and application security.
• Drafted and revised all company security policies, including those used by company counsel through development and production.
• Responsible for enforcement of policy and compliance standards applicable to the business.
• Front-end development of production code and dynamic AI feature development.
• Built with React / React Native, Java, Python, Node, and Flask, with extensive use of Linux.

Automation Toolkit

A desktop automation hub for an insurance brokerage — a system-tray app paired with a browser extension, consolidating form autofill, macro recording, and AI assistance into a single tool.

• Built the desktop application in Tauri with a Rust backend and a React + TypeScript frontend, paired with a Manifest V3 Chrome extension that bridges to the browser over a local WebSocket.
• Implemented Microsoft Entra sign-in using OAuth 2.0 with PKCE, storing tokens in the Windows Credential Manager rather than on disk, with silent refresh so users never re-authenticate.
• Shipped spreadsheet-driven form autofill, record/edit/replay browser macros, embedded AI chat, and a self-updating release channel.

Natural-Language Data Lake Access

An MCP server letting non-technical staff query the company data warehouse in plain English — without ever holding database credentials.

• Deployed a Model Context Protocol server on Cloud Run that translates natural-language questions into SQL against BigQuery and returns readable answers.
• Designed so no credentials live on user machines: authentication is centralized through a dedicated service account, keeping access auditable and revocable.
• Paired with a second MCP server providing read-only infrastructure visibility — service status, logs, and function inventory.

Producer Budget Application

A company-wide annual revenue-budgeting app that replaced a spreadsheet process with a guided form and validated submissions.

• Built as a Power Apps canvas app on Dataverse, with Power Automate flows handling data hydration from BigQuery and submission processing.
• Live-calculated metrics — recurring versus non-recurring growth, projected new business, budgeted revenue — update as users type, with per-account completion tracking.
• Submissions export as structured payloads and return to the user as a downloadable record.

Support Ticket Routing Analysis

An analysis of seven months of internal support tickets to determine how much of the triage queue could be automated — and where automation was not worth it.

• Analyzed resolved tickets to derive deterministic routing rules mapping ticket category to the correct agent.
• Found that a simple lookup table covered the majority of tickets at high accuracy, with no AI required.
• Recommended against adding an AI layer: it would have touched only a small ambiguous slice at meaningfully higher cost and complexity. Surfaced the remaining unowned categories as a decision for leadership.