August 24, 2026

Hiring Team
Atlas Public Services

Dear Hiring Team,

Atlas Public Services is working on the product question that has become central to my practice: how to use AI in an important service without obscuring risk or removing human judgment. I am applying for the Senior AI Product Manager role with nine years of product experience and direct responsibility for a bounded, human-reviewed AI pilot in case-management workflows.

At Harborlight Software, I led an AI-assisted case-summary pilot from scope and success measures through evaluation and rollout decisions. Final eligibility decisions remained with trained staff. I coordinated privacy review and frontline training, added opt-out controls and audit logs, and created an error taxonomy with reviewers. Weekly case review surfaced three recurring error patterns, and we stopped expansion until the highest-impact issue was resolved. The pilot reached 84% reviewer agreement on eligible cases and reduced manual triage effort by 18% without automating the final decision.

That work reinforced a lesson I would bring to Atlas: an aggregate metric is not enough unless the team understands whose outcome it represents and what action follows a failure. My broader product experience supports that approach. I lead discovery and roadmap outcomes for two cross-functional squads, partner with customer operations, design, engineering, privacy, and support, and have used service research and analytics to reduce median intake time by 31% and improve self-service completion from 62% to 78%. At Civic Grove Labs, I translated policy and service constraints into accessible public-interest products and phased partner releases.

My record supports privacy and frontline risk communication, but it does not explicitly establish legal partnership, federal procurement, or FedRAMP experience. I would be direct about those boundaries while bringing practical experience with evaluation, safeguards, escalation, and adoption. I would welcome a conversation about how Atlas defines quality and preserves human review across public-service AI workflows.

Sincerely,
Jordan Avery