Applied AI Engineer, Internal Systems

Minerva•New York City, NY

5h•Onsite

About The Position

Minerva is hiring an Applied AI Engineer to build and operate AI agents and automations for the company’s internal teams. The role serves as a forward-deployed engineer for Minerva, working with operations, finance, people, recruiting, sales and the founders to identify high-effort or high-risk processes and replace them with reliable systems that include audit trails. The position reports to the COO and works closely with engineering, which owns systems that touch production. The engineer will own a company knowledge agent spanning documentation, Slack, calls and tickets, with Slack as its interface. The role also automates internal workflows, manages a company-wide library of Claude-connected skills, establishes information sources of truth, and builds data plumbing for operational dashboards, controls and board reporting. Security and governance are central: systems should use least-privilege access, require human approval before agents write to systems of record, and maintain audit trails suitable as compliance evidence. The engineer works directly with internal users to understand their processes, measures adoption and model costs, and maintains runbooks for systems that remain in use. The role is based in person at Minerva’s Williamsburg, Brooklyn office. Candidates must be authorized to work in the United States; visa sponsorship is considered case by case.

Requirements

Has shipped agents or automations that coworkers use weekly and can demonstrate usage and measurable time savings.
Production experience with Python or TypeScript and connecting LLMs to real APIs.
Comfort with OAuth scopes, webhooks, scheduled jobs and their failure modes.
Understands company operations and can communicate effectively with recruiting, finance and security stakeholders.
Applies a controls mindset by deciding approval ownership and failure handling before shipping systems.
Keeps current with AI tooling and can demonstrate a personal setup.
Writes clearly and produces runbooks that users can follow independently.
Lives in New York City or is willing to relocate.
Must be authorized to work in the United States; visa sponsorship is considered case by case.
Applicants must provide an example of something they built and others used, such as a demo, repository or description of the process, usage and failures.

Nice To Haves

Has rolled out Claude or another model platform company-wide, including access, permissions, usage and cost management.
Has participated in a SOC 2 or similar audit on the evidence-producing side.
Hands-on experience with startup systems such as HRIS, ATS, CRM, billing and spend tools.

Responsibilities

Build and own a company knowledge agent over documents, Slack, calls and tickets, using Slack as the front end and providing sourced answers.
Automate internal company processes, including hiring, onboarding and offboarding, access provisioning, vendor reviews, customer security questionnaires, contract renewals and billing.
Manage the skill library, connecting Claude to team tools and ensuring each skill has a named team owner, an evaluation and appropriate permissions.
Define and maintain sources of truth for company information, making it accessible to people and agents and routing proposed updates for human approval.
Build the data plumbing for operational dashboards, metrics, controls and board reporting.
Establish guardrails, including least-privilege access, human approval before agent writes to systems of record, and audit trails that support compliance evidence.
Observe how internal users work and automate the processes they most want to eliminate.
Measure adoption, system usage, model costs and agent permissions; retire unused systems and document retained systems in runbooks.
Set a baseline and measure monthly hours of manual work retired.
Deliver the stated early outcomes, including a daily-used Slack knowledge agent, faster new-hire setup and at least three workflows that remain in use after four weeks.