Forward Deployed Engineer

Synphony · On-site with customers, heavy travel · Agriculture · Manufacturing · Mining · Oil & Gas

The job in one sentence: show up at a farm, a cable plant, a mine, or an oil field, find the expensive problem nobody's solved, and build the entire thing that solves it — data pipeline, model, agent, dashboard, integration — mostly by yourself.

What we are. Synphony is the deployment layer for physical AI. In plain terms: we take frontier AI — software systems, agents, and robots — and make it actually work inside the messy physical industries that run the real economy. We don't run demos. We deploy into heat, dust, legacy machines, and people who've done the job for thirty years and don't care about your model architecture. The company that owns the connective tissue between the floor, the line, and the org wins. That's the job.

The role. This is not a normal engineering job and the bar is not normal. A Forward Deployed Engineer is a one-person strike team. You get dropped into a customer with a vague, costly problem and you own it end to end — scope it with their engineers, wrangle their filthy data, build the system, ship it to production, and stand next to the machine when it runs. No tickets handed down, no clean spec, no team of ten to lean on. You and the problem.

Here's why the bar is brutal: in a single quarter you might build a predictive-quality model for a manufacturer, wire an ops agent into a grower's workflow, and stand up a sensor pipeline feeding robot training data — and we need that from one person, not three specialists. Generalists who actually ship win here. Specialists who need a clean lane don't.

What you'll actually build — some of these, on rotation, depending on the customer:

• Machine analytics & observability — ML on sensor and telemetry data: predictive maintenance, anomaly detection, quality prediction, scrap and COPQ models. Join filthy industrial records through inconsistent IDs, find the signal, ship the model, put it behind a dashboard an operator actually trusts.
• AI agents — email agents, ops agents, multi-agent frameworks that take real action inside a customer's workflow. LLM orchestration that survives contact with a real business instead of dying in a demo.
• ERP & org intelligence — integrate into the systems the business truly runs on, and make data flow between the floor and the people making decisions.
• Robotics data — build the pipelines that turn raw sensor and teleop data into clean training data for robot policies. Sensor integration and data plumbing — the unglamorous 80% that makes robot learning actually work.

The bar — you've probably done several of these:

• Shipped real ML/data systems to production solo — owned the pipeline, the model, and the deployment, not just a notebook.
• Built with LLM agents or multi-agent frameworks and made them do something real.
• Done hard data engineering: messy, multi-source, no schema, and you joined it anyway.
• Touched robotics, sensors, or hardware data and weren't scared of it.
• Stood up your own infra (cloud, k8s/k3s, whatever it took) without waiting for a platform team.
• Sat across from a customer and turned "this keeps breaking and costing us money" into a built, working solution.

You don't need all of it. You need range, raw building ability, and the kind of ownership where "that's not my area" isn't a sentence you say.

The honest part. You'll travel. You'll be on-site in places that are hot, loud, dusty, and far from a good coffee. You'll work with data that's a disaster and people who are skeptical of you until you earn it. You'll context-switch across industries and stacks constantly. If that sounds miserable, this isn't your job. If it sounds like the most fun you could have as an engineer, we should talk.

Why it's worth it. You'll have more surface area and more ownership than any normal role would ever hand you — at the customer site, you are the company. You'll build across the entire physical-AI stack instead of one narrow slice of it. And you'll be early at the company building the operating system for the industries that grow the food, dig the materials, and make the things — the dirty jobs nobody else will touch. That's the point.