Applied AI Engineer

The role

You own how Paires converses: every reply handled well and in time, booking, nurture, and the coaching experiences we are building on the same data. It starts as our reply and messaging system and grows into a core part of the product. You will also build broader applied-AI features across the platform. What this seat is not: a research seat. An internal knowledge bot or a RAG demo is not the bar. We want people whose systems hold real conversations with real people outside their team, day after day. Cold outreach from nothing is a different seat too - here the conversations already exist, and your agents carry them.

What you will own

• Agents that read inbound messages and draft the right reply, fast, with a human in the loop.
• The messaging layer across our warm outreach, our warm investor network, and investor relations, not just one inbox.
• Conversational agents end to end: replies, booking, nurture, and support that reads human.
• The eval spine that gates quality: golden sets, judges, and the guardrails that hold as the conversations grow.
• Broader applied-AI features: classification, extraction, routing, summarization.

You are a fit if you

• Ship conversational LLM systems to production with real users: reply handling, support, booking, or nurture agents. Inbound work counts fully here.
• Have built evals yourself: golden sets from scratch, judge criteria, ship or rollback decisions made on the numbers.
• Have integrated LLMs with email, CRM, or messaging systems.
• Think like an operator, and know the business goal behind the message.
• Are an engineer first. We run roughly 80/20 engineering to research.
• Move fast with AI tooling and own outcomes.
• Our stack: Python, Supabase, Pydantic AI, Claude Agent SDK, AWS. If you have shipped on any of it, lead with that. Bonus: email and CRM integrations, RevOps exposure, sales or IR experience.