Applied AI Engineer, Internal Systems

Description

The company builds AI agents that take over marketing work, including analysis, campaign planning, activation, measurement and reporting. It has raised $20M from venture investors. The team combines operators and investors from finance, advertising and tech with engineers and researchers from leading universities.

The stack centres on Claude, used company-wide in Slack, through the API, with skills and MCP. Agents the team builds use Mastra, and long-running work runs on Temporal. Code is Python and TypeScript, with scheduled jobs and webhooks. Day-to-day tools include Slack, Google Workspace, Notion, and the CRM, recruiting, HR, finance and compliance systems.

A company of this size carries a lot of unseen work: copying a new hire's details into several tools, tracking which contract renews next, and answering the same security questionnaire again. Most companies would hire a coordinator for it. This company is hiring an engineer.

This is a forward-deployed role, but the customer is the company itself: operations, finance, people, recruiting, sales and the founders. You work alongside these teams, find the tasks they most want to drop, and build a system that runs without them and keeps an audit trail. Some firms call this an AI operations engineer or internal AI engineer.

Backgrounds that fit include a founder or early engineer who automated the back office, a chief of staff or operations lead who learned to build, a solutions or forward-deployed engineer who wants the whole company as a customer, a GTM or internal tools engineer who wants scope past sales, or anyone who has automated their own job at least once.

This role is not IT helpdesk work, product engineering on customer-facing agents or production infrastructure, or an AI strategy job. It also does not own the work it automates: compliance owns security answers and sales ops owns the CRM.

In the first 90 days, the knowledge agent should be live in Slack and used daily, a new hire should be set up in every tool in under an hour, and at least three workflows should run without a person in the loop and still be in use four weeks later. In the first six months, security questionnaires should go out the same day without a founder writing them, renewals and price increases should surface before they cost money, and every team should have a system it would want to keep.

You own the hours of manual work retired each month, measured against a baseline you set in your first month.

Responsibilities

• Build and run a company-wide knowledge agent across docs, Slack, calls and tickets, with Slack as the interface. Its answers must cite sources so people trust them.
• Automate company operations: hiring, onboarding and offboarding, access provisioning, vendor reviews, customer security questionnaires, contract renewals and billing. Begin with the processes that consume the most hours or carry the most risk.
• Manage the skill library. Connect Claude to each team's tools and give each team a set of skills, each with a named owner on that team, an evaluation and correct permissions.
• Decide where each type of information lives, move it there, and keep it findable for people and agents. When an agent finds outdated information, it proposes a correction and a person approves it.
• Build the plumbing behind the operating dashboards, including metrics, controls and board reporting that are currently assembled by hand from five systems.
• Set guardrails first: least-privilege access, human approval before any agent writes to a system of record, and audit trails that also serve as compliance evidence.
• Observe how recruiters, sellers and the finance lead work, then automate the step they dislike most.
• Track adoption: which systems are used, by whom, and at what cost, including model spend and what each agent can access. Retire tools nobody uses and write runbooks for the rest.

Requirements

• You have shipped agents or automations that coworkers use every week, and you can show usage numbers: who stopped doing which task and how many hours were freed.
• You write production Python or TypeScript and have connected LLMs to live APIs. You understand OAuth scopes, webhooks and scheduled jobs, and how each one fails.
• You understand how companies operate and can discuss work with a recruiter, a controller or a security auditor in their terms. You fix the process before automating it.
• You think about controls. Before shipping, you have decided who approves what and what happens when an agent is wrong.
• You keep up with AI tooling and can show your own working setup.
• You write clearly, and people can follow your runbooks without asking you questions.
• You live in New York City or are willing to relocate.

Conditions

Base salary of $211,000 to $236,000, depending on experience, plus meaningful equity and medical, dental and vision coverage.

You report to the COO and work closely with engineering, which owns anything that touches production.

The role is fully in person at the office at 300 Kent Avenue in Williamsburg, Brooklyn.

Benefits

• Meaningful equity.
• Medical, dental and vision coverage.