Projects

Designed and deployed a fully on-premises, agency-controlled artificial intelligence system for the Alabama Board of Medical Examiners. Built on enterprise-grade hardware with vector search, document ingestion pipeline, and Active Directory integration — enabling secure, compliant AI capabilities with zero external API calls.

• Enterprise hardware: Dell Pro Max Tower T2 with Intel Core Ultra 9 285 (24 cores, 5.6 GHz), NVIDIA RTX PRO 6000 Blackwell (96GB GDDR7), 128GB DDR5 RAM, dual 2TB NVMe RAID 1.
• AI stack: Ollama (open-source LLM inference), Gemma 4 31B (Google's flagship model, 256K token context), Qdrant vector database (14,784 documents indexed, 152,000+ chunks with PII detection).
• Security & integration: HTTPS via self-signed certificates, Active Directory LDAP authentication, Open WebUI with role-based permissions, Nginx reverse proxy, full encryption in transit and at rest.
• Document ingestion: Custom Python pipeline with Docling (PDF/OCR), Microsoft Presidio (PII detection), nomic-embed-text vectorization, running 4 workers during off-hours, 1 during business hours, fully resumable via SQLite checkpoint.
• Regulatory compliance: Hard limits on probable cause determinations, charging decisions, adjudications. System prompt enforces BME AI Use Policy, OIT compliance, and work product attribution automatically.

Migrated decades of on-premises file server records to Azure and deployed a Retrieval Augmented Generation (RAG) solution enabling intelligent, citation-backed querying of legacy agency documents.

• Ingested on-premises file server content into Azure Blob Storage, building a structured document repository for AI indexing.
• Deployed Azure AI Search with an enrichment pipeline that chunks documents, generates embeddings, and builds a searchable index.
• Leveraged Azure OpenAI (GPT-4) to power a conversational interface that grounds responses in agency records with source citations.
• Enabled staff to query decades of legacy records, legal cases, and agency data in seconds — dramatically reducing manual research time.
• Built on the Microsoft PubSec Information Assistant open-source architecture, adapted for the agency's on-premises-to-cloud data migration.

Led end-to-end design and implementation of a VMware Horizon VDI environment on Dell VxRail for the Alabama Department of Examiners of Public Accounts, enabling secure remote work for 185 employees.

• Migrated 97% of users from physical desktops to a fully virtualized environment.
• Deployed across two vSphere environments plus one hyper-converged Dell VxRail cluster.
• Implemented network redundancy from ISP to Horizon server to ensure zero single points of failure.
• Reduced hardware costs and enabled seamless remote work capabilities statewide.

Planned and executed a full Microsoft 365 tenant migration for a state agency, separating their environment from another government organization with zero data loss.

• Migrated all email, SharePoint, OneDrive, and Teams data to a new sovereign tenant.
• Configured Azure AD, MFA, and SSO for all applications post-migration.
• Minimized downtime through careful cutover planning and staff communication.

Implemented a FortiVoice VM-200 VoIP phone system for the Alabama Board of Medical Examiners, replacing legacy hardware with a modern, software-defined solution.

• Deployed FortiVoice with soft clients for desktop and mobile use.
• Configured policy-based routing to ensure call reliability and redundancy.
• Provided staff training and documentation for the new system.

Designed and implemented a cloud-based disaster recovery solution to protect critical state data and ensure business continuity.

• Implemented AWS off-site backups integrated with Dell Data Domain for on-premises deduplication.
• Established recovery time and point objectives aligned with agency compliance requirements.
• Documented runbooks and tested recovery procedures to validate the solution.

Designed and shipped a fully functional, production-grade interactive resume website from scratch — with zero prior coding experience — using Claude AI as a collaborative development partner.

• Built with Next.js 14, TypeScript, and Tailwind CSS; deployed on Vercel with a custom domain.
• Integrated OpenAI gpt-4o-mini to power an 'Ask My Resume' AI chat widget trained on resume content.
• Implemented CI/CD workflow via GitHub pull requests with automatic Vercel deployments on merge.
• Featured as a live demonstration of AI-assisted development and modern IT infrastructure skills.

A Raspberry Pi-powered home automation and monitoring project inspired by the Marauder's Map from Harry Potter — a real-time dashboard that tracks activity and presence across the house, nicknamed 'the pumpkin house.'

• Built on Raspberry Pi with Arduino sensors for room-level presence and environmental monitoring.
• Designed to provide a live, map-style view of household activity — who's home, which rooms are active, and environmental conditions.
• Demonstrates hands-on IoT skills including hardware integration, sensor data collection, and home network configuration.
• Personal project applying enterprise infrastructure thinking (monitoring, alerting, uptime) to a home environment.