Forward Deployed Engineer

Responsibilities:

• Scope and build customer voice workflows on the canvas: pre-call lookups, in-call tools, knowledge bases, transfers, post-call data extraction, and webhooks.
• Write the integrations customers need (REST and webhook glue into CRMs, ticketing, scheduling, and internal systems) and make them robust to retries, timeouts, and bad data.
• Run onboarding end-to-end: discovery calls, prompt and workflow design, Simulation test passes, telephony number setup with the customer's carrier, go-live, and the first weeks of monitoring.
• Debug production calls using our observability stack: latency by STT / LLM / TTS component, end-status breakdowns, tool-call traces and recordings.
• Turn repeated customer asks into product: file precise issues, ship small platform fixes yourself (Python agent runtime, Go APIs), and hand larger ones to the core team with a reproduction.
• Write the runbook and hand-off notes so Customer Success can support the account without you.
• Be on point for your customers during launch windows, including some US-hours overlap for North American accounts.

Requirements:

• 1.5 to 4 years building and running software in production, in a product company or a fast-moving services team.
• Strong in Python; comfortable in TypeScript; willing to read and patch Go.
• Has integrated with third-party APIs for real (auth flows, pagination, webhooks, idempotency) and knows what breaks when they misbehave.
• Has worked with LLM APIs: prompting, structured outputs, tool/function calling, and evaluating whether a change actually helped.
• Comfortable with Docker, a cloud provider (we use AWS), and reading logs, metrics, and traces to find the cause of a bad call.
• Can sit in a customer call, understand a business process, and translate it into a workflow with clear edge cases, then explain trade-offs in plain language.
• Writes clearly: PRs, issues, runbooks, and customer-facing notes that someone else can act on.
• Bias to shipping. Comfortable owning an outcome with incomplete information and asking for help early.
• None of these are required. Voice experience is a plus, not a filter; we will teach you telephony.
• Any exposure to voice or real-time systems: LiveKit, WebRTC, SIP, Twilio / Telnyx / Exotel, STT or TTS APIs, streaming audio.
• Built or shipped an LLM agent with tools in production, and dealt with latency, hallucination, or cost trade-offs.
• Go in production, or Kubernetes beyond kubectl apply.
• Worked directly with customers before: solutions engineering, implementation, or a founding-engineer role at an early-stage company.
• Contact-centre, healthcare, or fintech domain knowledge (several of our customers are in these).
• Familiar with ClickHouse, Kafka, or similar analytics pipelines.