Forward Deployed AI & Product Engineer

You’ll work at the intersection of engineering, AI, product, customers, and go-to-market.

This is fundamentally an engineering role, but not one where the customer is abstracted behind a roadmap. You’ll develop our platform, work directly with prospects and customers, run technical demos, onboard new accounts, integrate their CRM and telephony, launch pilots, troubleshoot production deployments, and turn what you learn in the field into better product and infrastructure.

You help win the customer, build the platform, deploy the customer, and improve the product because of what you learned doing all three.

About us

We’re building an AI-native sales and communications platform spanning AI roleplay, realtime coaching, outbound sales, inbound voice agents, qualification, negotiation, follow-up, CRM activity, and operational workflows.

Freight brokerage is our beachhead, and we have an unfair advantage there: we own and operate a mid-sized freight brokerage. That means we build and prove the platform against a live business with real reps, real call volume, real CRM, and real revenue on the line, not a hypothetical customer. You’ll deploy into an operation we control before we take it to anyone else, so the product is battle-tested before it ever hits an outside account.

Freight exposes the problem at full intensity: high call volume, fragmented systems, repetitive qualification and negotiation, and workflows where AI must understand business context and take controlled action. The platform will not remain freight-specific.

Why this role exists

The quality of our product is inseparable from the quality of its engineering and deployment.

A voice agent can look excellent in a controlled demo and fail on a real PSTN call with interruptions, short utterances, latency, background noise, missing CRM context, and a customer expecting an immediate answer. A great product can also fail if the CRM mapping is wrong, telephony is misconfigured, ownership rules are misunderstood, or the deployed workflow does not match how the customer actually operates.

Those problems do not divide neatly into Engineering, Product, Sales Engineering, Implementation, and Customer Success tickets.

Someone has to follow the whole problem. That is this role.

What you’ll doDevelop the platform

This is a hands-on production engineering role. You will write, review, and merge production code across the same platform you demonstrate and deploy with customers.

Our application stack includes:

• Next.js 16, React 19, TypeScript, and Node.js
• Supabase and PostgreSQL, including RLS, Realtime, migrations, and multi-tenant data models
• Inngest for durable asynchronous workflows, retries, CRM synchronization, AI processing, and orchestration
• Electron for live-call and realtime coaching experiences
• Vercel, GitHub, and CI/CD for deployment and production operations
• REST APIs, WebSockets, authentication, authorization, observability, and distributed application infrastructure

You should be comfortable moving from frontend product work to backend logic, databases, integrations, asynchronous systems, and production debugging without needing a tour guide.

You may build a customer-facing workflow in the morning, modify CRM synchronization in the afternoon, and trace a realtime voice defect later that day. You own problems through the stack rather than stopping at the boundary of a team or service.

Build realtime voice AI

Voice is a core part of the product.

You will work with PSTN calling, inbound and outbound DIDs, media streams, WebSockets, call routing, recordings, transcripts, live coaching, warm transfers, voicemail, interruption handling, and realtime audio.

Our architecture is deliberately provider-flexible. Depending on the capability and evidence, you may work with or evaluate OpenAI Realtime, Hume EVI, Deepgram, ElevenLabs, Groq, PersonaPlex, Parakeet/NVIDIA Riva, self-hosted ASR/TTS, and emerging speech-to-speech models.

You should understand architectures such as:

Audio → VAD / Turn Detection → ASR → LLM / Reasoning → Tools & Business Context → TTS → Streaming Playback

as well as direct speech-to-speech systems.

You’ll reason about endpointing, partial transcripts, streaming, barge-in, cancellation, audio buffering, first-audio latency, short-turn behavior, PSTN degradation, and model-versus-infrastructure failures.

Treat AI quality as an engineering problem

“Sounds good” is not an acceptance…