AI Engineer — Customer Support Copilot

Samuel Torres

Full-time · Mid-level · Toronto

This role is no longer accepting applications

Similar open roles you can apply to now:

About the role

I lead this team, so I will say it plainly: the copilot works, but not well enough yet. It drafts useful replies about 70 percent of the time, and that last 30 percent is exactly why we are hiring. Support leaders do not care that an answer is fluent if it is wrong, slow, or too generic to send. We need someone who likes tightening systems, not demoing them. The job is a mix of retrieval work, prompt and context design, evaluation, and product judgment about where AI should stop and a human should step in. If you have built LLM features that real customers use every day, you already know the kind of trade-offs we mean.

Responsibilities

Improve draft reply and summarisation quality for our support copilot across email and chat workflows
Own retrieval improvements over our help centre and resolved ticket corpus
Build offline and online evaluation loops so prompt and pipeline changes are measured rather than guessed
Reduce latency and token cost while maintaining answer quality for high-volume customers
Work with product managers and design to decide where automation helps and where human review must remain primary

Requirements

3+ years of software engineering with at least 1 year building LLM-backed product features in production
Python for backend services and evaluation tooling
LangChain or equivalent orchestration framework for multi-step retrieval and generation flows
OpenAI API experience including structured outputs, tool calling, and streaming responses
RAG experience with retrieval tuning, chunking strategy, and source quality evaluation
SQL for analysing ticket data, conversation patterns, and feature usage outcomes