Full Stack Engineer

Full Stack Engineer — Raleigh, NC
· In-office
· Full-time

Hey — Scot here, CEO and co-founder of Re Fi Buy . If you're the kind of engineer who reads release notes for fun, ships on day one, has been quietly building agents on the side, and gets a little restless on teams that talk more than they build → keep reading.

• No recruiters, please.
• Applicants must have permanent work authorization in the US. We're not sponsoring visas for this role.
• This is an in-office role in Raleigh, NC. If you're not in the Triangle, you'd need to relocate at your own expense— we don't have a relocation budget for this seat.
• You'll work daily with our Dublin engineering team. Comfort with async + a couple of early/late calls a week is part of the job.

You’ll help us build an AI-native, agentic e‑commerce platform end‑to‑end — shipping features across a Next.js / Type Script front end and a C# / .NET 9 back end on .NET Aspire that powers an event‑driven, multi‑tenant SaaS platform. You’ll spend as much time wiring up production agent workflows (enrichment, scoring, multi‑engine monitoring) as you do building UI. We want builders who own the whole feature — schema to UI to telemetry to rollout.

Today we’re running tens of thousands of SKUs through agentic enrichment, distributing optimized catalogs to engines like ChatGPT, Perplexity, Gemini, and Copilot, and tracking visibility back from each one daily. The work is real, the customers are real, and the agents are in production.

• Frontend:
  Next.js, React, Type Script, Tailwind/shadcn, SSR/ISR, server components, plus a Chrome extension surface.
• Backend: C# / .NET 9 on .NET Aspire, microservices, minimal APIs, DI, background workers, feature‑flagged rollouts.
• Eventing:
  Event‑driven Dapr architecture with pub/sub.
• AI‑native:
  Auto Gen, Semantic Kernel, frontier LLMs (OpenAI/Anthropic/Google), retrieval, function/tool calling, prompt+response evals, agent orchestration in production.
• Dev Ex:
  Git Hub monorepo, Git Hub Actions, trunk‑based dev with feature flags, Linear for cycles + tickets.
• Observability:
  Open Telemetry tracing/metrics/logs, dashboards, and a growing fleet of internal Claude‑powered ops agents.

• Build product experiences in Next.js + Type Script with polished UX.
• Design resilient .NET / Aspire services and event‑driven workflows — idempotency, retries, DLQs, the unglamorous stuff that keeps agents and feeds honest at 100k+ SKU scale.
• Ship and own production LLM agent processes: catalog enrichment, attribute generation, Q&A automation, and the Visibility / Quality / Digital Shelf scoring loop.
• Build the multi‑engine monitoring layer — measuring presence and quality across ChatGPT, Perplexity, Claude, META, Gemini, and Copilot.
• Wire up agentic ops: extend the agents that run feed health, PR reviews, backlog audits, and prod monitoring.
• Work on Reinforcement Learning with AI Feedback Loops as your main gig, not a hobby.
• Own features end‑to‑end → schema/design, API contracts, evals + tests, telemetry, feature‑flag rollout, learnings.
• Contribute to platform and devex: CI/CD, preview envs, AI coding tools, internal libraries.
• 3+ years shipping production apps with modern Type Script/React or C#/.NET.
• Solid grasp of SSR, state management, API design, and relational data modeling.
• Hands‑on experience with at least one LLM API and an agent or RAG pattern — even side projects count, but you've actually wired up tool calling and seen what breaks.
• Comfort with containers, cloud deploys, Git, code review hygiene, and automated testing fundamentals.

• .NET 9 + .NET Aspire, minimal APIs, background jobs, Clean Architecture, CQRS.
• Event‑driven messaging with Dapr (or Kafka/NATS/SNS+SQS).
• Production agent work:
  Auto Gen, Semantic Kernel, Lang Chain, or your own framework. Bonus for evals, guardrails, and cost/latency tuning.
• Vector DBs, RAG, hybrid retrieval, response evaluation.
• Ecommerce: catalogs, marketplaces (Amazon/Walmart/eBay), feeds, attribute taxonomy, retailer integrations.
• Linear / cycle‑based planning, feature‑flag‑driven release engineering, on‑call comfort.

Builder culture → fast iterations, small PRs,…