Founding AI Engineer

Paires; Canada; On-site; CAD 120,000 - 190,000; Full time

Benefits offered by this job

Fully remote

Equity potential

Job summary

Paires is hiring a Founding AI Engineer to own the agent layer and the matching engine that powers investor outreach and relationship management. You will build, scale, and run the core AI surface, shaping how founders connect with investors.

We require hands-on experience shipping production AI systems, strong Python skills, and a track record with embeddings or retrieval. This is a fully remote role with equity potential for the right person.

Qualifications

• Shipping production software with real-world impact.
• Experience deploying LLM-based systems in production.
• Strong Python fluency and data/system thinking.
• Experience with embeddings, ranking or retrieval.
• Ability to own end-to-end product surface, from ideation to launch.

Responsibilities

• Own the agent layer and the matching engine end-to-end.
• Scale and maintain production-grade AI components.
• Design evals to ensure high-quality outputs.
• Collaborate with a flat team and ship fast.

Skills

Python

LLM systems

Production software

Embeddings / matching

FastAPI

Tools

Claude Agent SDK

Pydantic AI

PostgreSQL

Supabase

AWS

FastAPI

Job description

We are hiring our Founding AI Engineer to own the agent layer of an AI-first fundraising platform: the conversational agents that run our warm outreach and investor relationships, and the matching engine behind them.

Paires is where founders come to raise capital. We pair them with the right investors from a large, engaged global investor network, then our agents run the warm outreach and manage the relationships that turn into meetings. It is a two-sided platform, a product for founders and a living network on the investor side, not one-way matching. We are live with paying clients, profitable and self-funded, and we run as a small, senior, flat team that ships fast.

The role

You own our core: the agent system that runs the product - agents that research investors, draft outreach that reads human, and carry investor conversations end to end - plus the matching engine that pairs founders with investors behind them. You build it, you scale it, and you run it. This is the most important surface in the product, and it is yours.

What you will own

• The agent layer: one orchestrator routing work to specialist agents. Those agents research investors, draft outreach that reads human, and carry investor conversations end to end. Evals gate what they are allowed to do alone. This is where you start.

• The matching engine behind it: embeddings, ranking, scoring, and the feedback loop that makes matches sharper over time.

• The backend and data behind it all: Python, Postgres, Supabase, AWS, clean pipelines. The agent layer is built on the Claude Agent SDK and Pydantic AI.

• Shipping to production end to end, and the judgment to know what to build next as we scale.

You are a fit if you

• Have been one of the first engineers at a quick-growing company, running the show: you know what it takes to scale and deliver a product, not just manage it.

• Have at least 3 years of shipping production software behind you, and have taken a full app from zero to production and kept it running, end to end. This seat needs a whole-product builder, not only a model specialist.

• Have shipped LLM systems to production and have real examples you can walk us through, including an agent system that is more than a single prompt in a loop.

• Have built retrieval, ranking, or matching with embeddings in production.

• Write strong Python and reason clearly about data and systems.

• Design evals, because you care whether the output is actually good.

• Own outcomes end to end and move fast with AI tooling. You write the code yourself: this is a builder seat, not a management seat. There is no team to hand work to.

• Bonus: TypeScript in production, FastAPI, vector databases, fintech or fundraising.

What we offer

Fully remote and async. Your day overlaps with US Eastern time for a few hours - not full US hours. Meetings batch on Mondays and Thursdays, the rest is deep work. The best AI tooling, paid (Claude Code, Cursor, top models). Real ownership of the core product, with equity potential for the right person.