Software Engineer (AI Agents)

Fluent is building the next generation advertising network: Partner Monetize and Advertiser Acquisition. Our vision is to build an ML/AI first network of advertisers and publishers working toward a common objective, elevating relevancy in e-commerce for everyday shoppers.

Fluent is hiring its first dedicated AI Engineer. Over the past year, AI at Fluent has moved from pilots to production: a service desk agent that closes tickets without a human touch, a pipeline that cut modeled audience delivery from days to hours, and an agent that audits cross-system attribution and identifies revenue reconciliation opportunities. All of it was built by people doing it alongside their day jobs. Your job is to make it your day job.

You will sit in Core Engineering and work every day with Data Engineering, Data Science, and Solution Architecture. You will build the agents and automated workflows our business teams have asked for, and you will build them on the Databricks lakehouse that is Fluent's data hub, because the hard part of most of these workflows is not the model. It is getting trusted data and reliable actions in front of it.

You are a builder who ships. You are strong in Python and SQL, you have put at least one LLM-powered application in front of real users, and you know the difference between a demo and a production system: evaluation, monitoring, cost, and a clear owner. You have worked in Databricks or a comparable lakehouse, or you have the data engineering depth to get productive there quickly. You are comfortable sitting with a support lead or an account manager, understanding how they actually work, and turning that into something that runs every day. This role is fully remote in the United States, with occasional travel to our New York headquarters.

What you'll do

• Ship production agents and automated workflows from a prioritized, business-validated backlog. Own each one from scoping with the business owner through build, deployment, monitoring, documentation, and handoff.
• Build the data foundation those workflows depend on in Databricks: Unity Catalog tables and views, Lakeflow pipelines, Lakeflow Connect ingestion from systems like HubSpot, and Genie spaces that let business teams ask questions in plain language. Bring data from our ad-serving and partner platforms, CRM, support desk, and finance systems into one place people can trust.
• Work with Data Science to make agent quality measurable: evaluation sets, accuracy and drift tracking, cost and latency instrumentation, and the audit fields and failure alerts every production workflow carries.
• Integrate with Fluent's systems through APIs, webhooks, MCP servers, and orchestration (Databricks Workflows, n8n, GitHub Actions), and replace brittle point-to-point automations with maintained, observable ones.
• Partner with Core Engineering to expose platform data and actions as APIs, so agents and business teams can use them directly instead of routing requests through engineers.
• Build reusable pieces, not one-offs: Claude skills, MCP connectors, prompt and evaluation libraries, and playbooks that other builders across Fluent can pick up.
• Handle consumer data in line with Fluent's privacy and compliance standards, use only approved AI tools, and work with IT & Security on credentials and access.
• Demo weekly, write things down in Confluence, and share what you learn in brown bags and tech talks.
• Use AI-assisted development (Claude Code, GitHub Copilot review) as a normal part of how you work.

What success looks like

In your first 90 days you will have shipped one workflow end to end, with a named business owner and numbers Finance can validate. By six months you will have several in production, at least one reusable component another builder has picked up, and a clear view of what next quarter's backlog should be.