Build the sales engineering layer for dvLED.

Most companies scale customer success by adding people. We scale it with agents, and you design them. Our customers are dvLED manufacturers and distributors whose sales teams configure, quote and draw LED walls in Arterio. You own their success after the signature, and you turn the work around it, from support and reporting to execution, into agentic workflows you deploy, supervise and keep improving. Agents carry the volume. You carry the judgment and the relationships.

Details

• Reports to the CEO, Irvine, California
• Works daily with engineering in Berlin
• Start as soon as we find each other

What you will do

• Build the support layer: an agent that answers reps where they work, in the CRM, in Slack and by email, grounded in each customer's catalog, price books and rules, and hands the hard cases to you or engineering with the full context attached.
• Automate reporting: account health and adoption for us; cabinets, pitches, regions and costs for our customers. Delivered by agents that explain what changed and why.
• Turn onboarding into a pipeline: agents read a manufacturer's datasheets and price lists, load cabinets, price books and engineering rules, and validate them against the engines. You sign off on the go-live.
• Run the account motion with agents: follow-ups, CRM hygiene, renewal prep and expansion into new regions, seats and modules, with a human in the loop where it counts.
• Treat every workflow as a product: write the instructions, set the guardrails, build evals from real tickets and quotes, and measure what your agents resolve, how fast and how well.
• Own what agents cannot: train sales reps and sales engineers, build each customer's playbook for configuring on the call, and show up on site and at the trade shows where this industry meets.

What you bring

• Five or more years in customer success, account management or sales engineering for B2B software or pro AV.
• Hands-on with AI: you have built agents or automations other people rely on, and you know where models fail and how to design around it.
• Fluency with data and systems: price books, product specs, CRM objects, APIs. You can read a wiring plan after we show you one.
• A clear writer, which counts twice here: customers read what you write, and so do your agents.
• Familiarity with dvLED or pro AV (cabinets, processing, mounting, drawings), or the appetite to learn it in your first quarter.
• Based in the United States, with roughly 20 percent travel.

Nice to have

• Admin-level Salesforce or HubSpot, including their AI agents.
• Enough Python, SQL or TypeScript to connect an API or query a database yourself.
• Time at an LED manufacturer, distributor or integrator.
• German, for the daily thread with the team in Berlin.

We are building agentic dvLED, and you build the road there. Arterio's engines already turn a manufacturer's product data into the complete package: wall geometry, processing, data and power wiring, pricing, drawings and documents. Now we open them to agents, step by step: an MCP server first, then fully autonomous configuration, then integration across the processing, mounts, CRMs and agent platforms our customers already run. The model brings the reasoning. The engines make every answer engineering-grade.

Details

• Reports to the CTO, Berlin
• Small team, direct ownership, your code in front of customers and their agents within weeks
• Start as soon as we find each other

What you will do

• Build the Arterio MCP server: every engine a typed, well-described tool, scoped to each customer's catalog, rules and permissions, so any agent can configure a wall, price it and pull the drawings.
• Push toward fully autonomous configuration: an agent takes an RFQ, a floor plan or a two-line email and returns the complete package. Agents propose, and the engines verify every port, mount, circuit and price.
• Build the evals: real projects as test cases, scored on correctness, latency and cost, so every model, prompt and tool change ships with evidence.
• Integrate the ecosystem: processing platforms, universal mounts, Salesforce and HubSpot, and the agent platforms our customers adopt, so a configuration starts and finishes wherever the work happens.
• Keep the human product sharp: the configurator interface and the quotes, wiring plans and dimensioned drawings it generates as PDF and DXF, instant while the customer is watching.
• Model real hardware as code that sales engineers and agents both trust: port capacities, mount constraints, circuit loads, clearances.

What you bring

• Four or more years building production web applications, front end and back end.
• Strong TypeScript and a modern web framework, and comfort in SQL. You care about types because the output is engineering-grade.
• Real experience building with LLMs: tool calling, structured outputs, retrieval and evals. You have shipped something where a model calls your code, and you know how it breaks.
• A taste for computational problems: geometry, path planning, constraint checks, and the arithmetic of prices and tiers.
• You build with coding agents every day, and you still read the datasheet before you trust the function.
• Based in the United States, with morning overlap with Berlin.

Nice to have

• An MCP server or other agent tooling running in production.
• Generating vector output: SVG, DXF, PDF.
• CRM APIs, Salesforce in particular, or any CPQ system.
• Pro AV, hardware configuration or CAD tooling in your past.