Forward Deployed Engineer

The role

Help customers turn a promising AI workflow into something they can trust in production. At Orchestra, that starts with understanding the task: what the model must get right, what failure costs, and where a better prompt, route, or specialist could make a difference.

You will work across customer discovery, integration, evaluation, and deployment. The job is to ship useful improvements, then bring what you learn back into the product so the next customer can move faster.

What you’ll do

• Work directly with customer engineers and domain experts to choose a concrete workload and define its quality, latency, and cost targets.
• Integrate model calls, tools, and production traces into a workflow that can be inspected and evaluated.
• Turn examples and expert corrections into task-specific checks, with held-out work kept separate from optimization.
• Improve prompts, output schemas, and tool contracts; compare model routes; collaborate on specialist training when the task warrants it.
• Take successful experiments into production with a clear rollout plan, fallbacks, and operational visibility.
• Diagnose failures across the application, gateway, and model boundary, and turn recurring integration work into reusable product capabilities.

What you bring

• Strong software engineering skills and experience taking integrations or applications from an unclear starting point into production.
• Comfort working with APIs, data pipelines, and the messy details of real customer systems.
• Practical experience building with language models, including evaluation and structured or tool-driven workflows.
• The ability to ask good questions, explain technical tradeoffs, and work alongside people who know their domain better than you do.
• Sound judgment about evidence: a cheaper request is not automatically a better outcome, and a successful API response is not proof that the task was completed.
• Ownership of the whole result, including the debugging and follow-through after a demo works.

Tell us what you’ve built.

Tell us about a system you shipped for someone else. What did the user need, what made the work difficult, and how did you know it worked? Code, a demo, a technical write-up, or a clear account of your contribution are all useful. Please leave out confidential customer information.

Opens your email app. You can also write to founders@understudylabs.com.