Senior Full Stack Engineer
Own and ship core systems that let AI agents reliably drive material flows, routing, and reconciliation at scale.
Take the challenge
We hire this role by what you build, not by where you've worked. Skip the resume. Open the challenge, use any tools you want, and show us how you'd run the work.
About the role
Replenysh is building the infrastructure for circular supply chains — connecting brands to a 5,000+ partner network with automated routing, real-time custody tracking, policy enforcement, and audit-ready proof. AI agents will drive more of the day-to-day work. We need a full-stack builder who can create the data models, agent tooling, integrations, and reliability layers that make this possible without creating operational debt.
Why this role matters
Real supply chain data is messy. When AI agents start making routing decisions and reconciling documents, weak foundations turn into constant exceptions. A strong builder who can own both the backend systems and the interfaces and tools around them creates leverage instead of fragility.
What you will own
- —Core data models and state for orders, material flows, custody events, and provenance that work for both humans and AI agents.
- —Routing and matching foundations — primitives, scoring, guardrails — that support reliable agent-driven decisions.
- —Event and tracking systems that agents can monitor and act on in real time.
- —Integration and ingestion layers that handle noisy real-world signals (documents, partner data) while keeping everything auditable.
- —Agent tooling and interfaces that make automated workflows observable and controllable.
- —Reliability and reconciliation primitives that keep agent-driven operations stable as volume grows.
Who we are looking for
- —Strong full-stack builder with experience shipping production systems at scale.
- —AI-native — you use Claude Code, or your AI tool of choice, as a serious collaborator and design systems that agents can reliably use.
- —High-agency: you see messy operational problems, clarify the real invariants, and ship working software.
- —Thrives in nuance: detail-oriented and rigorous with edge cases, data inconsistencies, and operational realities, while holding a clear vision for the full system and long-term leverage.
- —Comfortable owning both backend foundations and the surrounding tools and interfaces.
- —Thinks about auditability, guardrails, and long-term maintainability when agents are in the loop.
- —You use AI to leverage your knowledge and understand what you are shipping: the pros, the cons, and the edge cases.
- —You use data to refine your process and help your teammates improve.
How we will evaluate the challenge
- —Prioritization and the leverage you created — did you build the thing that matters most?
- —Clarity of your design decisions around state, identity, guardrails, and agent reliability.
- —How fluently and effectively you used AI tools.
- —How well the output handles real messiness while staying practical to run in production.
The challenge
It is your first month. New partners and rising order volume have left the data fragmented. You have a messy partner list, a pile of orders typed by humans, and a folder of weight tickets and BOLs in every format imaginable. Pick one agent to build and show us how you think.