Forward Deployed AI Architect

Alzheimer's Association®•Houston, TX

5d•$155,000 - $168,000•Remote

About The Position

The Forward Deployed AI Architect sits with a team, learns how their work actually happens, and then rebuilds that work with them using AI. Reporting to the Sr. Director of AI + Digital Transformation, with a close relationship to the Direct, AI Platform Operations, this role takes prioritized processes from across the Alzheimer’s Association and makes them agentic, in whole or in part. Each engagement is a sprint. Weeks to a few months, one process at a time, sometimes reaching across several business areas. The work starts with sitting next to people and watching what they actually do, at a level of detail needed for true transformation. It ends with something running that gives hours back, documentation the team can live with, and a clean handoff to AI Platform Operations and IT. This is a builder role. You will spend your days inside our AI platforms (primarily Claude) shipping working solutions rather than recommending them. When we need something outside our existing technology stack, you will say so and find what that is. Every hour this role gives back to the Association and our staff goes somewhere meaningful. Toward care, toward being in community, toward the research that ends this disease—toward the work only a person can do.

Requirements

Bachelor’s degree in a related field, or equivalent professional experience
3–5 years of professional experience in work that required understanding how a process actually runs and then changing it. Consulting, operations, product, program delivery, or a dedicated AI transformation role all qualify
Hands-on experience building AI agents, automations, or tools that real people adopted and continued using in their day-to-day work. Candidates should be prepared to walk through what they built, who used it, and what changed as a result
Working fluency with Claude or a comparable enterprise AI platform, including experience going beyond chat into agentic and multi-step workflows. Candidates coming from applied AI, forward deployed, or solutions delivery backgrounds are encouraged to apply
Comfort working with APIs, connectors, and structured data well enough to wire a solution to the systems where work lives
A clear understanding of what separates a prototype from a production system, and the judgment to know which one a given situation calls for
Ability to teach while building, so that the people you build for can maintain and extend the work after you leave
Strong written and verbal communication with non-technical staff, and the patience to work through a process with someone who is skeptical of AI
Comfort operating in a role with no established playbook, in an organization where trust is earned over time

Nice To Haves

Direct experience with Claude Code, Claude Cowork, MCP servers, or comparable agent development frameworks
Experience delivering AI or automation work inside a large, matrixed organization
Background in one or more of the areas the Association works in: scientific research, public policy and advocacy, healthcare, fundraising, or field and volunteer operations
Familiarity with Microsoft 365, Salesforce, Workday, or CXone as the systems a solution connects to
Prior nonprofit sector experience
Experience documenting and scaling a solution beyond the team it was built for

Responsibilities

Embed with a team and learn their work in granular detail. Sit through the actual task. Map where the time goes, where errors creep in, and which parts of the job people dread. Bring the team into the redesign from the first day so they own what comes out of it.
Build working agents, automations, skills, and tools using approved AI platforms, primarily Claude. Ship something people can use inside the sprint. Work in real time alongside the staff who will use it, iterating on their actual deliverables rather than a sample.
Run each engagement as a time-boxed sprint with a defined start and end. Set the value target up front. Prove it at the close. Hand the solution to AI Platform Operations and IT with the documentation, test cases, and runbook they need to support it, then stay available through a 30-day stabilization window before moving to the next process.
Diagnose and fix what you built while a sprint is live and through stabilization. After that, partner with the standing IT support group on resolution rather than owning it. Feed every failure back into how the next thing gets built.
Recognize when a solution built for one team fits four others. Turn one-off builds into reusable skills, agents, and templates. Maintain a working library that the AI Collective and staff across the Association can pull from without asking permission.
Define a measurable target for every engagement before the work starts, whether that is hours returned, cycle time reduced, or capacity created. Track it through delivery and after. Report results to the Sr. Director and the AI Steering Committee, including the engagements that did not pay off.
Work inside the Association’s Responsible AI guidelines, data handling standards, and information security requirements. Know which processes should stay in human hands and say so early. Partner with Enterprise Risk and AI Platform Operations before any solution touches constituent, donor, research, or staff data.
Maintain a close working relationship with the Director of AI Platform Operations, who owns platform configuration, enterprise connectors, security review, and consumption. Coordinate early when a build needs new system access or new credentials. Serve as a dedicated member of the AI Collective.