Product Engineer, Agent Economics

You’ll own the economics of a protocol where AI agents negotiate early-payment terms on real invoices - and much more

We’re building agent-to-agent negotiation for invoice early payment: when a buyer approves an invoice on Causa Prima, the supplier’s agent, the buyer’s agent, and potentially competing underwriters enter a sealed-bid, second-price auction that clears in seconds. The cryptography keeps bids private. The mechanism has to make truthful behavior the smart play for every participant — suppliers, buyers, financiers, and us.

That second part is your job. You’ll own the economic and market design of this protocol: how it’s priced, how it’s calibrated, who participates, and why every party keeps showing up. You’ll work directly with the founders and our Head of Product, at the intersection of mechanism design, supply chain finance, and agentic AI.

What you’ll do

• Calibrate the mechanism. Build simulations of auction outcomes against synthetic supplier and buyer populations, measure efficiency loss against first-best, and turn results into shipped design decisions — not papers.

• Govern pricing policy. Design the signed, versioned, deterministic pricing policies our in-house underwriting book bids with — sophisticated enough to price risk, rule-bound enough to be externally defensible.

• Shape policy expression. Own how suppliers and buyers express their thresholds and other parameters: defaults pre-seeded from accounting data, the review-and-override experience, and how a cleared APR translates into terms an SMB owner actually understands.

• Guard incentive integrity. Fee structure, fair-rate index, disclosure framing, and the structural separation between running the venue and bidding in it. If suppliers ever conclude we’re a buyer-side extraction tool, the network fails — your job is to make that structurally impossible.

• Stress-test adversarially. Information leakage in repeated play, collusion vectors, prompt-injection surfaces in agent decision functions. You ask how the system gets gamed before asking how it scales.

What we’re looking for

• Mechanism design depth. Deep grounding in auction theory, market design, or mechanism design — academic or applied — and the judgment to know where classical assumptions stop holding when the players are AI agents, not rational humans.

• Financial services fluency. Experience in supply chain finance, payments, credit, or treasury. You know what an APR means to a CFO and what dilution risk means to a financier.

• An adversarial mind. Threat modeling is instinct, not checklist. You’ve broken systems — or proven why they couldn’t be broken.

• Hands-on agentic AI experience. You’ve built or rigorously tested LLM-based agent workflows yourself — with Claude Code, Cursor, or similar tools — not just read about them.

• Bias to shipped decisions. You turn ambiguity into a default proposal, an owner, and a deadline. Comfort being measured on live deployment metrics, not research output.

• Fluent English. German or Spanish is a plus.

Nice to have

• Applied auction work. Sealed-bid auctions, MPC-based markets, or token-economic mechanism verification in production or near-production settings.

• Receivables finance context. Familiarity with dynamic discounting, reverse factoring, or multi-financier marketplace models (Taulia, CRX, TReDS-style), including SCF accounting treatment.

• Published work. Research or open-source contributions in mechanism design, market design, or agentic systems.