Digital Product Manager / Systems builder

Product thinking.
Systems fluency.

I'm Andrew Gallagher. I turn complex insurance operations into practical digital workflows, and keep my technical skills grounded by building and running my own infrastructure.

Product perspective. Hands-on curiosity.

Explore the professional work and the lab: two ways I turn complex systems into useful experiences.

View my résumé ↗Explore professional workExplore the AI sysadmin project
SYSTEMS / CONNECTED THINKINGExplore a relationship ↗
Product strategy Priorities + people Data & APIs Reusable signals Business needs Operational context Automation Connected workflows Applied AI Guardrails + utility Infrastructure Hands-on systems
CONNECTION IN FOCUS Product strategy

Company-wide planning across approximately 50 stakeholders connects business priorities, technical constraints, and delivery.

Hover or focus a node to preview its story; activate it to explore.

~12,000Notices handled annuallyConnected rules, deadlines, and workflows
~20 → ~1 minNotice-identification runtimeA database process I optimized
~50Stakeholders across departments12 planning meetings every three weeks
Explore all professional case studies →

01 / Professional work

Complex operations.
Practical improvements.

Selected work at McNeil & Co. My contribution, the result, and the decisions behind it.

Workflow modernization

From manual notices to connected workflows

Helped connect qualification rules, deadlines, assignments, and delivery for a process handling approximately 12,000 notices a year.

Identification runtime: ~20 minutes to ~1 minute

SQL ServerAPIs / PostmanWorkflow design
Read the case study: From manual notices to connected workflows

The problem

Reports, spreadsheets, manual assignments, and repeated copy/paste made state-dependent notice work fragmented.

My contribution

Helped design and evolve the rules, database logic, assignment workflows, and deadline calculations. Tested vendor APIs with developers and helped connect review, sending, filing, and printing. Optimized the notice-identification database process.

The outcome

A connected notice process at approximately 12,000 notices annually, with a key identification process reduced from roughly 20 minutes to about 1 minute.

The decision behind it

Treat accuracy, deadlines, and the work between systems as part of the product. A faster query matters when it enables a more usable operational workflow.

Platform strategy

One portal, different people and needs

Helped shape a shared, role-based portal strategy for agents, insureds, and other external audiences.

Shared capabilities with role-specific experiences

Portal strategyUser experienceStakeholder alignment
Read the case study: One portal, different people and needs

The problem

Separate audience-specific portals duplicated functionality and made change harder.

My contribution

Helped drive a unified, role-based approach and advocated for magic-link authentication for people who use the portal infrequently.

The outcome

A strategy centered on shared capabilities and audience-specific access. Adoption and user-volume metrics have not been quantified.

The decision behind it

Start with how often someone uses the product and what they need to accomplish. Infrequent users should not have to fight the sign-in experience.

Applied AI

Turning AI interest into useful work

Lead the AI user group and help teams mature practical use cases, including consolidated claims summaries.

A repeatable path from use case to governance review

AI enablementCopilotGovernance
Read the case study: Turning AI interest into useful work

The problem

Teams needed a way to develop AI ideas into useful applications. Claims users also had to piece together history, timelines, and financial details across records.

My contribution

Proposed and lead the AI user group, helped shape a complementary governance structure, and present AI topics internally. Helped enable a Copilot workflow that brings claim history, timeline, and financial information into a structured summary.

The outcome

An organizational mechanism for developing AI use cases and a consolidated claims review surface. Adoption and time savings remain unquantified.

The decision behind it

Begin with a concrete workflow and the people doing it. Give ideas room to mature before governance review, and keep the source information central to the experience.

Hands-on problem solving

The tool that opened the door to IT

Built an Excel/VBA pricing calculator that made cross-line pricing and modification factors easier to explore.

Pricing modeled accurately within cents

Excel / VBARating logicOperational tools
Read the case study: The tool that opened the door to IT

The problem

Pricing adjustments involved repeated guess-and-check rating changes with limited visibility across lines of business.

My contribution

Built a calculator with interactive controls and modification-factor modeling. Developed advanced VBA tooling using classes, database connectivity, custom ribbon controls, and reusable add-ins.

The outcome

The calculator modeled pricing within cents. This process-improvement work drew IT attention and helped lead to my move into a Business Analyst role.

The decision behind it

Understanding the operation made it possible to build a useful tool. Making the logic reusable and deployable extended the value beyond a single worksheet.

Product leadership

Keeping a company-wide backlog connected

Run roughly 12 planning meetings every three weeks across approximately 50 stakeholders and multiple departments.

Shared priorities, clearer intent, visible dependencies

PrioritizationStory definitionCross-functional delivery
Read the case study: Keeping a company-wide backlog connected

The problem

Company-wide work spans underwriting, claims, finance, HR, and risk management, with competing priorities and dependencies.

My contribution

Shape backlog work and stories, facilitate planning, clarify business intent, and contribute to scope, UAT, and release discussions. Work with business leadership and the CIO on shared priorities.

The outcome

A recurring planning cadence that supports structured delivery and root-cause problem solving. Improvements in IT perception and ticket volume are team contributions.

The decision behind it

A meeting is useful when it resolves intent, dependencies, or a decision. My role is to make business context actionable for the people doing the work.

Data & automation

Making underwriting information reusable

Worked on API-driven property enrichment and advocated for policy-form rules built from data already in the system.

Structured data for analysis and repeatable decisions

API integrationData modelingRules-driven workflows
Read the case study: Making underwriting information reusable

The problem

Property information lived in documents and manual lookups, while policy forms were manually selected despite existing coverage and product data.

My contribution

Worked on address-based API retrieval and persistence of property fields, including year built and construction characteristics. Pushed stakeholders to define rules for selecting required policy forms.

The outcome

More reusable exposure data for underwriting analysis and rules that reduce manual form checking. Quantitative error and throughput measurements remain to be established.

The decision behind it

Capture information as structured data once, then use it across the workflow. Existing data often provides the starting point for a more consistent process.

02 / The homelab

Curiosity, put into practice.

Real infrastructure, real changes, and the troubleshooting and recovery decisions behind them.

Interactive architecture / Not live telemetry

One lab. Several layers.

Four compute nodes connected by deliberate choices about access, identity, observability and recovery. Select a node to inspect its role.

Edge & accessVPS · Traefik · Pangolin

Controlled entry points and remote service access

IdentityAuthentik · OIDC

A reusable login layer for supported services

ObservabilityPrometheus · Grafana · Uptime Kuma

Separate service health signals from assumptions

RecoveryProxmox Backup Server · console-access plan

Recovery options considered before consequential maintenance

Compute / Select a node

GAIA / Intel NUC

Sierra

Operations and lightweight services

A compact Proxmox node for always-on utility workloads, monitoring, dashboards and supporting services.

Core responsibilities
  • Service monitoring
  • Lightweight workloads
  • Dashboards and automation
Systems lesson

Right-size workloads to the hardware; reliability matters more than novelty.

GAIA / NAS / ZFS

K2

Storage and data services

The storage-focused node runs ZFS-backed services and supports data-heavy work in the lab.

Core responsibilities
  • Proxmox
  • ZFS storage
  • Storage operations
Systems lesson

Data integrity and recovery planning belong in the architecture from the beginning.

GAIA / Dell workstation

Fuji

Media and home automation

Hosts services including Jellyfin and a Home Assistant OS virtual machine, connecting personal applications with physical devices.

Read the related project →
Core responsibilities
  • Virtual machines
  • Home Assistant
  • Containerized services
Systems lesson

The interface people rely on should remain understandable even when the implementation spans multiple services.

GAIA / NVIDIA RTX 3080 Mobile / 64 GB RAM

Olympus

GPU and local inference

Provides GPU-backed local inference with Ollama in an LXC container. Preserving acceleration across system changes was a central part of the Proxmox upgrade.

Read the related project →
Core responsibilities
  • NVIDIA / CUDA
  • Ollama in LXC
  • Host-container diagnostics
Systems lesson

A visible GPU is not the same as a correctly accelerated workload: verify the behavior across every layer.

Architectural illustration only. No live host state, internal URLs, credentials or privileged configuration are exposed.

02

Gaia: a self-hosted ecosystem

A multi-node homelab bringing together compute, storage, identity, remote access, and services I use at home.

ProxmoxLinux / LXCDockerZFS

What I work with

Distributing services across hardware with different capabilities; running containers and virtual machines; connecting local services through Traefik and remote access through a VPS.

What this demonstrates

Systems integration, Linux administration, networking, and practical decisions about where a service belongs.

Inside the project

The lab combines a NAS, an Intel NUC, a Dell system, and a GPU-equipped laptop. Authentik provides identity for supported services. Prometheus, Grafana, and Uptime Kuma provide monitoring, while Cloudflare DNS and certificate automation support service access.

03

Local AI on Olympus

Using an existing GPU laptop as a host for local language models, with Ollama running inside a Linux container.

OllamaNVIDIA / CUDALXCOpen WebUI

What I work with

GPU access inside a container, Linux driver compatibility, and the relationship between available hardware and local model choices.

What this demonstrates

Diagnosing issues across hardware, the host operating system, and the container environment.

Inside the project

Olympus has an RTX 3080 Mobile GPU and 64 GB of system memory. The Ollama container uses NVIDIA libraries and CUDA, with troubleshooting extending from GPU visibility to CUDA offload and actual model acceleration. Maintaining this workload has become a key part of my AI sysadmin project.

04

A home that responds

Connecting physical controls, lighting, and software through Home Assistant, with an emphasis on useful behavior.

Home AssistantZigbee / ZHAInovelliAutomation

What I work with

Smart-bulb mode, switch events, and lighting scenes that connect a familiar wall control to configurable behavior.

What this demonstrates

Translating a household need into an interaction, then troubleshooting the device and software layers that make it work.

Explore the Home Assistant project

05

One identity across services

Connecting supported self-hosted applications to Authentik, including a working OpenID Connect integration with Gitea and Google sign-in.

AuthentikOpenID ConnectGiteaIdentity & access

What I built

A shared identity layer for supported services, with groups organized around the people who use the lab and the access they need.

What this demonstrates

Integrating application authentication, understanding the relationship between sign-in and authorization, and making access decisions across a self-hosted environment.

Inside the project

Authentik connects Google sign-in to supported applications. I configured Gitea to use OpenID Connect and organized identity groups for family, friends, administrators, and different service roles.

03 / Practical experience

Connecting the parts.
Understanding the whole.

At work, I investigate data and integrations directly and translate business needs into delivery context. In the lab, I own the permissions, dependencies, and recovery path. Each strengthens the way I work with technical teams.

Systems & networking

Linux, Proxmox VE, Proxmox Backup Server, containers, GPU access, storage, DNS, and remote access.

Identity & observability

Authentik, OIDC integrations, Prometheus, Grafana, and uptime monitoring.

AI & automation

Scoped agent access, AI-assisted cluster upgrades, Ollama, Open WebUI, and n8n workflows.

Data & product tools

SQL Server, Postman, advanced Excel/VBA, story definition, prioritization, and Jira backlog management.

04 / How I got here

I learned the operation.
Then started improving it.

My path into product combines an early foundation in technical support with years inside insurance operations and underwriting. Building tools to solve the problems in front of me led to IT, business systems analysis, and company-wide digital product management.

That background still shapes my approach: learn how the work actually happens, identify the friction, and connect the people and systems needed to improve it. Outside work, the homelab keeps that learning practical.

Digital Product Manager

McNeil & Co. · Digital Product Manager since March 2020

I work across underwriting, claims, finance, HR, and risk management, shaping stories and priorities with business leadership and IT. My leadership comes through facilitation, technical credibility, and a clear understanding of the operation.

I lead the AI user group, present at IT town halls, and work directly with developers, analysts, and vendors to turn product intent into workable decisions.

The technical foundation

Long before managing digital products, I supported campus users, repaired customer and business computers, and assisted with systems implementations. That experience taught me to diagnose problems firsthand and make technology understandable.

My education combines a bachelor's degree in Business Administration from the University at Buffalo with an Associate of Science in General Studies from Tompkins Cortland Community College. See my full career history →

A career built from the inside out

Full experience and résumé →

  1. 01 / Apr 2016 – Feb 2017Policy Analyst

    Policy issuance, forms, coverage, and underwriting data.

  2. 02 / Feb 2017 – Aug 2019Underwriting Assistant

    Rating, endorsements, proposals, and the pricing calculator.

  3. 03 / Aug 2019 – Mar 2020Business Analyst

    Core-system expertise, database logic, integration, and workflow automation.

  4. 04 / Mar 2020 – PresentDigital Product Manager

    Company-wide planning, modernization, stakeholder alignment, and applied AI.

Working on a complex operational product?

Let's talk about your team, the workflow you're improving, and where product judgment and technical fluency could help.

Start a conversation