AI Sales Associate

About the Role:

Most roles at the start of a career come with an onboarding deck and a defined task list. This one does not. Groupon is rebuilding how it sells — from the ground up, with AI — and the person in this role is part of that build from day one.

The vehicle is Project Foundry: a production fleet of AI agents designed to give Groupon a parallel sales force that operates 24/7 and to make every human rep sharper the moment they step into a deal. Two layers: the first runs full sales motions autonomously — outbound, inbound triage, reactivation — so reps receive warm opportunities rather than cold lists. The second equips reps at the handoff point — context surfaced, deal history ready, next action suggested.

Reporting directly to the CSO, the AI Sales Associate learns the full architecture, contributes to both layers, and progressively takes end-to-end ownership of agents in production. You are not joining a team that advises the business on AI. You are building the system that is the business. There is no playbook. You help write it.

North Star

Ship AI agents that run on real merchant data, prove their value in measurable terms, and make the sales organisation faster and sharper than it would be without them. The build compounds — every signal, every call, every conversion feeds back into what you build next. You are here to make that happen.

What You’ll Do:

• Build across both layers of the fleet — Layer 1: agents that run autonomously without rep involvement — outbound sequencing, lead prioritisation, inbound triage, reactivation. Reps receive warm opportunities, not cold lists. Layer 2: agents that equip reps at the moment they step in — account context surfaced, deal history ready, next action suggested. You learn how both layers work and progressively contribute to each.
• Identify what to build and make the case for it — You are not waiting to be given a task list. You look at the sales organisation, find the highest-value problem an agent could solve, and bring a structured proposal to the CSO. Judgment about what is worth building matters as much as the ability to build it.
• Test, measure, and iterate — You run experiments on real merchant data, measure whether what you built is working, identify failure modes, and refine. Every agent you touch has a documented performance trail. You do not ship and move on.
• Mine call transcripts for agent inputs — Process sales call recordings to extract patterns, category signals, and performance data. This is raw material for the agents you build — not a separate analytics workstream. You surface insights from it and design around them.
• Document and maintain what you ship — Every agent you deploy has a before/after record. You keep it current. You present it. You own what you built — not just while you're building it.

What You Bring:

• A degree from an Ivy League or equivalent top-tier university — computer science, data science, economics, mathematics, or a field that taught you to think in systems. The subject matters less than the rigor of how you were trained to think.
• You have built an AI agent or automated workflow — not a class project. Something you designed, built, and iterated on because you wanted to see if it would work. You can explain what it did, what broke, and what you changed. The bar is not commercial success. The bar is that you shipped something real.
• Technical capability to build: you have used LLM API